Rotary Drum Filter Selection: Precoat and Cake Wash

Last updated: 2026-11-06

A rotary drum filter is usually the right continuous vacuum filter for leach residues and precipitates once a duty needs 40–120 m² of active area; at a design flux of 500 kg/(m²·h), 120 m² (dry basis) handles roughly 60 t/h of solids. This guide covers sizing, precoat selection, cake washing, moisture troubleshooting, and running cost. For the wider equipment landscape, begin at our solid-liquid separation hub.

Key Takeaways

• Size from leaf-test data: design flux typically 300–900 kg/(m²·h), to be confirmed by test work.
• Use precoat for gelatinous hydroxides and slimes; expect 20–80 kg/h of filter aid consumption.
• Above roughly 50 m² per drum, add parallel units instead of one very large machine.
• Cake moisture of 8–25 wt% is normal; climbing moisture almost always has a mechanical cause.
• Budget 6–15 kWh per dry tonne of solids, with the vacuum pump dominating the draw.

What Is a Rotary Drum Filter and Where Does It Sit in a Flowsheet?

A rotary drum filter is a continuous vacuum filter that forms, washes, dries, and discharges a filter cake on a slowly rotating, partially submerged cylindrical drum. The drum surface is divided into shallow cells, each covered with filter cloth, and each cell connects through a rotary valve to the vacuum, wash, and blow systems in sequence. Because the cycle repeats every revolution, the machine delivers a steady cake stream without batch handling.

Five terms recur throughout this article, and getting them precise prevents most specification errors:

• Submergence is the arc fraction of the drum that sits below the slurry surface, typically 25–45%, and it sets how much of each revolution is available for cake formation.
• Form time is the portion of each revolution that a cell spends under vacuum while its cloth is on the slurry, drawing liquor through the growing cake.
• Precoat is a thick sacrificial layer of filter aid that is built up on the drum before slurry is introduced, then shaved away — together with the trapped solids — by an advancing knife.
• Cake wash is the displacement wash that replaces mother liquor trapped in the cake with water or process liquor, applied by weirs or spray headers after cake formation and before the dry zone.
• Discharge is the means of removing the cake — scraper knife, strings, rollers, or a leaving cloth — each with different solids and cloth-life consequences.

In a hydrometallurgical flowsheet, the machine normally sits downstream of thickening: see our notes on high-rate thickener selection for the underflow that feeds it, typically 40–55 wt% solids. Typical duties include leach residues, ferric hydroxide from iron removal, nickel and cobalt precipitates, gypsum, and red mud. Where the slurry is viscous or carries long-chain flocculant, check slurry rheology first, because a pumpable, non-segregating slurry is a precondition for any vacuum filter. For lab confirmation of cake permeability and wash behavior, a lab-scale rotary drum vacuum filter reproduces form time and submergence faithfully at bench scale.

Which Parameters Set Rotary Drum Filter Capacity?

Capacity is never a single number; it is the product of available area, effective cycle time, and achievable flux. Table 1 lists the parameters a process engineer controls or must specify, with typical ranges.

Every value in Table 1 is a typical range, to be confirmed by test work on your slurry. Two parameters interact more than the rest. Drum speed and cake thickness trade against each other through form time: the cake deposited per revolution scales roughly with the form time, so doubling speed halves thickness at constant flux.

For coarse, free-draining gypsum, a fast drum with a thin cake is preferred because flux improves with shorter cycles while moisture barely suffers. For fine, slow-settling precipitates, a slow drum with a deeper cake is preferred because the cake must be thick enough to support its own washing without cracking.

Precoat or Bare Cloth: When Does the Duty Need Filter Aid?

A bare-cloth machine works when the solids form a permeable, non-blinding cake: gypsum, jarosite-type precipitates, crystallized salts, and cleanly sized residue sands. A precoat drum filter becomes unavoidable when the solids are gelatinous or slime-graded: ferric hydroxide from iron removal, nickel and cobalt hydroxides, red mud fines, and colloidal silica. On these feeds, bare cloth blinds within one to three shifts, and no amount of back-pulsing recovers the lost flux.

The price of precoat is real and must appear in the operating budget:

• Filter aid consumption of 20–80 kg/h on a mid-size unit, driven by knife advance and the solids loading captured in the aid layer.
• Cake purity changes, because the first few millimeters of every cake contain traces of diatomaceous earth or expanded perlite — acceptable for a residue, a specification problem if the cake is a saleable product.
• Knife discipline: the blade must advance 0.05–0.2 mm per revolution, and a dull or misaligned knife wastes precoat and smears fines back into the layer.
• Cloth life, ironically, improves under precoat because the cloth never touches the slurry directly.

Quantify the trade-off with specific cake resistance data before committing either way. For gelatinous hydroxide precipitates, a precoat drum filter is preferred because the alternative is cloth replacement every few days and a filtrate clarity you cannot hold to specification.

How Do You Size a Rotary Drum Filter for 60 t/h of Solids?

The worked example below uses 60 t/h (66 short t/h) of dry solids on a residue filtering at a design flux of 500 kg/(m²·h). All figures are dry basis and exclude the standby unit unless noted.

1. Define the feed. 45 wt% solids after thickening, so the slurry to the filters is 60 ÷ 0.45 ≈ 133 t/h. Record particle size, settling rate, temperature, and liquor density.
2. Run leaf tests. Test at the intended submergence, vacuum, and temperature, at three drum speeds. Our filter press sizing guide shows how bench leaf data translate into full-scale area.
3. Fix the design flux with margin. Apply a 0.7–0.8 factor to the best leaf-test flux to cover feed swings and scale-up; here we take 500 kg/(m²·h).
4. Compute the required area. 60,000 kg/h ÷ 500 kg/(m²·h) = 120 m² of active filtration area.
5. Split the area into units. Specify 4 × 30 m², three duty and one standby (spared installation). Standby coverage also absorbs the wash-out time each drum needs.
6. Check cake discharge. At 500 kg/(m²·h), a wet-cake bulk density near 1.3 t/m³, and 0.6 rpm, the deposit per revolution is roughly 11 mm — workable for a scraper, marginal enough to verify in test work.
7. Verify the auxiliaries. Size the vacuum pump, filtrate receivers, wash water system, and blow compressor against the worst case, not the average case.

Table 2 closes the mass balance for the example:

One judgment only surfaces after a few installations: a single-drum machine stops scaling gracefully at roughly 40–50 m². Beyond that you are fighting valve-port velocities, wear across ever-wider seal bars, and scraper knives several meters long that must track true over the full face. When the duty needs more area, adding parallel 25–40 m² drums with one shared standby is more robust — mechanically and commercially — than chasing one heroic machine. The same logic applies when comparing continuous vacuum options against alternatives such as a vacuum belt filter: parallel mid-size drums tolerate one unit’s maintenance without stopping the plant.

Why Does Cake Moisture Climb on a Rotary Drum Filter?

Climbing moisture is a symptom, not a disease. The seven faults below cover the great majority of cases we diagnose; each follows the pattern of what you observe, what causes it, and what to do.

1. Cloth blinding. Observation: moisture creeps up over weeks while vacuum holds. Cause: fines lodged in the weave, often cemented by scale or organics. Action: chemical cloth wash, verify back-blow, then re-cloth. The filter cloth blinding guide covers the mechanisms in depth.
2. Vacuum leaks. Observation: moisture high on all cakes, filtrate pump surging, receiver level erratic. Cause: rotary valve faces worn, gland seals leaking, cracked sight glasses. Action: smoke or ultrasonic test under vacuum, re-lap valve faces, replace seals.
3. Feed becomes finer. Observation: step change in moisture after an upstream change. Cause: finer grind or faster precipitation raises cake specific resistance, so the same cycle leaves a wetter cake. Action: re-run leaf tests, drop the flux assumption, coordinate with upstream on grind or pH trajectory.
4. Precoat exhausted or knife dull. Observation: rising moisture together with cloudy filtrate under precoat operation. Cause: the knife has stopped cutting into fresh aid, so the loaded layer no longer drains. Action: rebuild the precoat, check knife advance rate and blade edge.
5. Weak blow. Observation: a wet heel remains on the cloth at the six o’clock position. Cause: blow pressure below 0.3 bar(g) at the cell — often the compressor is fine but the rotary valve timing has drifted. Action: measure at the cell, not the compressor, and reset the valve bridge.
6. Wrong drum speed. Observation: thin, wet cakes (too fast) or thick, cracked cakes pulling air (too slow). Cause: speed moved outside the range the cake can support. Action: step-test across 0.1–1.5 rpm and log moisture against each setting.
7. Wash water short-circuiting. Observation: wash efficiency drops while moisture looks acceptable. Cause: liquor bypasses through cracks in the cake or floods off a tilted wash weir. Action: level the weir, reduce application rate, or switch to a gentler flood wash; a cracked cake needs low-head application, not more water.

Track the trend, not the spot value: a filter cake moisture log against drum speed, vacuum, and feed density will usually isolate the fault within a few shifts.

Which Discharge Arrangement Should You Specify?

Discharge determines what the machine tolerates, how long the cloth lives, and how much the unit costs. The four arrangements differ as follows:

• Scraper discharge. The default. A knife lifts the cake at the six o’clock position, aided by a blow pulse. It needs a cake of at least 10 mm, leaves a thin heel that eventually blinds, and gradually wears the cloth. Lowest capital and simplest maintenance.
• String discharge. Monofilament strings spiral the full drum width and lift the cake off before the knife. It suits sticky, fibrous, or flocculated cakes that a knife would smear, and it is gentle on the cloth. Cost: string tensioning and occasional re-threading.
• Roller discharge. A counter-rotating roll takes the cake off the cloth as a thin, sticky layer. It is the answer for cakes under 5 mm — pigments, fine hydroxides — that a knife cannot lift without gouging. Cost: the roll surface must be re-dressed or re-covered periodically.
• Cloth (belt) discharge. The cloth leaves the drum with the cake and returns over wash sprays on both sides. It gives the best cloth life, handles thin cakes, and keeps flux high on blinding duties. Cost: the highest capital of the four, plus cloth tracking hardware.

When the cake is under 5 mm and sticky, roller discharge is preferred because a scraper knife cannot lift a cake that thin without cutting into the cloth. When the duty is abrasive residue at 20–40 mm, scraper discharge is preferred because the thicker cake pays for the knife’s simplicity and the heel effect is negligible.

Specify the arrangement together with the cloth, not after it: string discharge needs a weave that anchors the monofilaments, roller discharge needs a reliably tacky cake, and cloth discharge needs tracking hardware sized for the drum width. Ordering cloth and discharge hardware from the same leaf-test campaign avoids the classic retrofit where a knife machine is re-clothed three times before it discharges cleanly.

What Does a Rotary Drum Filter Cost to Run?

Four items dominate. First, the vacuum system: expect 0.3–1.0 kW per m² of drum surface for the pump set, depending on vacuum level and seal-water design. Second, total electrical demand of 6–15 kWh per tonne of dry solids filtered (dry basis, duty units only, excluding tax, buildings, and standby depreciation) — on an internal estimate for a 120 m² installation with 55 kPa vacuum, we modeled about 9–11 kWh/t. Third, consumables: filter aid for precoat machines at the 20–80 kg/h noted above, plus cloth replacement on the 1,500–6,000 h cycle from Table 1. Fourth, labor: a continuous drum typically absorbs 0.2–0.5 of an operator’s time, more if precoat rebuilds are frequent.

Give any cost number an honest basis. State whether it is dry or wet basis, whether the standby unit is included, and whether tax and freight are in or out — otherwise two quotes can differ by 30% and both be “correct.”

Three Operating Snapshots (Indicative, Synthetic-Typical Basis)

The cases below are anonymized composites drawn from typical plant data; they are indicative, synthetic-typical basis, not guarantees.

Snapshot 1 — nickel intermediate precipitate. An anonymized 500 t/d plant (synthetic-typical basis) filtering a cobalt hydroxide stream on a 34 m² precoat drum: feed 9 t/h of dry solids (dry basis, single duty unit), flux 265 kg/(m²·h), filter aid 38 kg/h, knife advance 0.09 mm/rev, cake moisture 27 wt%. Filtrate ran clear to the tail of each precoat cycle. Similar duties appear throughout nickel HPAL flowsheets.

Snapshot 2 — zinc iron-removal residue. Two 20 m² bare-cloth drums on goethite residue: vacuum 55 kPa, drum 0.35 rpm, cake 18 mm, moisture 18 wt%, cloth life 4,100 h with weekly acid wash. The second drum covers wash-outs; effective availability 96% (excluding the standby unit, dry basis).

Snapshot 3 — copper precipitate wash duty. A 25 m² cloth-discharge drum washing a copper hydroxide cake: 2.5 m³ wash water per tonne of cake, single-stage cake wash efficiency 93% (soluble-salt basis), drum 0.8 rpm, cake 9 mm. Cloth discharge was chosen specifically because thin cakes blinded scraper machines within a shift.

FAQ

What flux should I use for sizing a rotary drum filter?
Start from leaf tests run at the design submergence, vacuum, and temperature. Typical fluxes span 300–900 kg/(m²·h): coarse gypsum near the top, gelatinous hydroxides near the bottom. Apply a 0.7–0.8 factor to the best leaf result to cover scale-up and feed swings. Never accept a vendor flux figure without the particle-size distribution and cake resistance data behind it; two slurries with similar names can differ by a factor of three.

How much filter aid does a precoat drum filter consume?
Consumption follows knife advance times the drum’s aid volume. For a 30 m² drum at 0.6 rpm and 0.1 mm/rev advance, expect roughly 25–45 kg/h of diatomaceous earth or perlite, inside the general 20–80 kg/h band (dry basis, duty unit only). At 50–100 mm of precoat depth, that gives one to four weeks between rebuilds, so rebuild labor belongs in the operating cost alongside the aid itself.

Can a rotary drum filter reach 10% cake moisture?
Only with coarse, crystalline solids and a healthy vacuum system. Gypsum and washed crystalline salts commonly reach 10–15 wt% under 60 kPa vacuum with a full dry zone. Gelatinous hydroxides and red mud rarely fall below 22–28 wt% regardless of effort; squeezing moisture further usually requires pressure filtration or steam drying, both of which change the economics of the whole flowsheet.

Does cake wash water dilute the filtrate circuit?
Yes, and the dilution is predictable: at 1–3 m³/t cake on the example duty, wash adds 75–225 m³/h of flow. Most plants split the filtrate system into a strong first filtrate that returns to leach and a weaker wash filtrate routed to wash-water makeup or effluent treatment. Design the receivers for two compartments from the start; retrofitting a split later means draining and re-piping a live filter train.

How long does filter cloth last on one of these machines?
Between 1,500 and 6,000 hours (typical range, to be confirmed by test work). The main drivers are abrasive content, blinding chemistry — scale, organics, or gypsum reprecipitation in the weave — and discharge type. Cloth discharge lasts longest because both faces are washed every revolution; scraper discharge is hardest because the knife wears the cloth surface slightly with every pass and a residual heel bakes onto it between blows.

When is a disc filter or belt filter the better choice?
A disc filter wins when you need very large area in a small footprint for coarse, easily filtered solids that need no wash — its vertical elements cannot carry a wash stage. A horizontal belt filter wins on fast-settling solids needing thorough washing with clean separation of mother and wash liquor. The drum sits between: it washes, tolerates thinner cakes than belts on some duties, and fits existing buildings. Compare the options in our vacuum versus pressure filtration overview before fixing the topology.

What test work do I need before ordering?
At minimum: leaf tests at three speeds and two vacuum levels, a cake wash efficiency test at two water rates, slurry viscosity and settling data, and a full liquor analysis including scale-forming species. If precoat is on the table, add a bench precoat trial to fix knife advance and aid selection. Two hundred liters of well-mixed, representative slurry supports all of it; a month of engineering time saved by skipping the test work is usually repaid by one wrong machine.

If your flowsheet includes a residue or precipitate filtration duty, send the feed rate, particle size, temperature, and liquor chemistry through our request a quote page — a process engineer, not a salesperson, will size the machine and tell you honestly whether precoat is justified.

Conclusion

The rotary drum filter earns its place in hydrometallurgy by being the compromise that keeps working: it filters continuously, washes a vertical cake without mother-and-wash liquor mixing hardware, and scales through parallel units long before a single drum reaches its practical limit. Get the sizing from leaf-test flux rather than catalog optimism, spend filter aid only where the solids genuinely demand it, and treat climbing moisture as a mechanical fault with a short list of causes. Do those three things and the machine will run quietly for decades.

References

• U.S. Geological Survey, Mineral Commodity Summaries. https://www.usgs.gov/
• SME Mineral Processing and Extractive Metallurgy Handbook. Society for Mining, Metallurgy & Exploration.
• Wills’ Mineral Processing Technology, 8th edition. Butterworth-Heinemann.
• Hydrometallurgy conference proceedings, TMS / MetSoc of CIM.
• CSIRO Mineral Resources, filtration and dewatering research publications. https://www.csiro.au/

Changelog

• 2026-11-06: Initial publication.

Related equipment

• 80mm Rotary Drum Centrifugal Extractor — Centrifugal Extractors
• 50mm Rotary Drum Centrifugal Extractor — Centrifugal Extractors