A Stoneware Filter is a fluid clarification device built around one or more porous ceramic (vitrified stoneware) candles housed in a pressure vessel. It traps suspended particulates as fluid passes from outside the candle wall to a clean inner core, removing solids down to 0.5 µm without chemicals or membranes. We use it where fine particulate must come out of low-viscosity fluids — process water, dilute acids, brewery wort polish — and where membrane filters would foul too fast. A typical 7-candle housing handles 2-4 m³/h of clarified flow before back-flushing.
Stoneware Filter Interactive Calculator
Vary candle count and per-candle flux range to see the total clarified flow capacity and choke margin for a stoneware candle filter.
Equation Used
This calculator scales the per-candle clarified flow range by the number of installed stoneware candles. The article states that a typical 7-candle housing handles about 2-4 m3/h before back-flushing, so the default per-candle low and high flows are set to 2000/7 and 4000/7 L/h.
- Flow capacity scales linearly with the number of identical stoneware candles.
- Per-candle low and high flow values represent the expected operating range before back-flushing.
- A margin is shown relative to the article warning that flux below 100 L/h per candle can choke the housing.
Operating Principle of the Stoneware Filter
The working element is a hollow cylindrical candle made from kaolin and feldspar, kiln-fired at around 1200 °C until the body vitrifies into a rigid stoneware matrix with a controlled pore network of roughly 0.5 to 5 µm. Fluid enters the housing under pressure, typically 1-4 bar, and flows radially through the candle wall. Particulates larger than the pore rating get caught at the outer surface and inside the first millimetre of the wall thickness, while clarified fluid collects in the central bore and exits through the top manifold. This is dead-end depth filtration with a surface-cake component — the same physics behind a Berkefeld filter or the older Pasteur-Chamberland candles, just scaled up for industrial duty.
Why stoneware and not sintered metal or polymer? Stoneware tolerates aggressive chemistry. It shrugs off dilute sulphuric acid, hot caustic in short cycles, and steam sterilisation at 125 °C without dimensional change. The vitrified body resists abrasion from sand and scale particles that would chew up a wound polypropylene cartridge in a week. The catch is the candle is brittle — a careless wrench drop cracks it, and thermal shock above roughly 80 °C/min splits the wall along the firing grain.
If the pore rating drifts during firing, you lose either flow or filtration. A candle fired too cold runs porous, lets 10 µm particles through, and ruins downstream polish. Fired too hot, pores close, flux drops below 100 L/h per candle, and the housing chokes within hours. Common failure modes are pinhole cracks from water-hammer (install a slow-opening valve upstream), gradual pore blinding by colloidal iron or organics (back-flush with reverse air at 2 bar every shift), and gasket extrusion at the top flange when operators over-torque the candle nut past 25 Nm.
Key Components
- Stoneware filter candle: The hollow vitrified ceramic element doing the actual filtration. Standard sizes run 250-1000 mm long with 50-70 mm outside diameter and a 20-30 mm inner bore. Pore rating is set during firing and cannot be adjusted in the field — specify 0.5, 1, 3, or 5 µm at order time.
- Pressure housing: Carbon steel or 316L stainless vessel rated to 6 bar working pressure. Holds 1 to 24 candles in a tube-sheet pattern. The housing must drain fully — even a 50 mL low-spot pocket grows biofilm during downtime and inoculates the next batch.
- Top tube-sheet and gaskets: Locates each candle and seals the dirty side from the clean side. Gaskets are typically EPDM or PTFE-jacketed silicone. Gasket compression must hit 30-40% — under-compressed and you get bypass, over-compressed and the candle neck fractures.
- Differential pressure gauge: Reads ΔP across the candle bank, normally 0.2-0.5 bar clean. When ΔP hits 1.5 bar, back-flush. When it hits 2.5 bar after back-flush, the candles are blinded and need acid clean or replacement.
- Back-flush manifold: Reverses clean-side air or filtrate through the candles at 2-3 bar to dislodge surface cake. A typical cycle is 30 seconds reverse flow every 4-8 hours of forward filtration.
Where the Stoneware Filter Is Used
Stoneware filters earn their place wherever the fluid is chemically nasty, the particulates are fine, and uptime matters more than absolute flux. They suit batch and semi-continuous duty where you can afford a 2-minute back-flush every few hours. You see them most in chemical process trains, beverage clarification, plating shops, and pharmaceutical pre-filtration ahead of cartridge polishers. They are not the right answer for very high flux water plants — that's where multimedia or membrane filters win — but for nasty intermediates carrying 50-500 mg/L of suspended solids, they outlast every alternative.
- Chemical processing: Polishing dilute sulphuric acid pickle liquor at a Tata Steel cold-rolling line in Jamshedpur, where stoneware candles strip iron-hydroxide flocs ahead of acid regeneration.
- Brewing: Bright-beer polish at the Pilsner Urquell brewery in Plzeň, using 1 µm stoneware candles after kieselguhr filtration to catch yeast and DE carry-over before the BBT.
- Pharmaceutical: Pre-filtration of fermentation broth at a Lonza API plant in Visp, Switzerland, removing mycelial debris ahead of a 0.2 µm sterilising cartridge.
- Electroplating: Continuous polishing of nickel-sulphamate plating bath at an MDA Aerospace job-shop in Brampton, Ontario, where 3 µm candles trap anode sludge that pits plated parts.
- Potable water: Point-of-entry sediment removal at remote lodges — a Berkefeld-style 4-candle gravity unit handles surface-water turbidity at the Patagonia Camp lodge near Torres del Paine.
- Edible oil refining: Polishing bleached palm oil at a Wilmar refinery in Pasir Gudang, Malaysia, with stoneware candles pre-coated in bleaching earth to drop colour and trace metals.
The Formula Behind the Stoneware Filter
The number that sets every stoneware filter design is filtrate flux — how many litres per hour you get per square metre of candle surface at a given driving pressure. At the low end of the typical range, 50 L/(h·m²) at 0.5 bar ΔP, you're filtering a heavy slurry or a viscous oil and the housing will spend most of its day in back-flush. At the nominal sweet spot, around 200 L/(h·m²) at 1 bar, candles run 4-8 hours between cleans and total candle life stretches past 2 years. Push above 500 L/(h·m²) by cranking ΔP past 2 bar and you'll fracture candles within weeks — the wall sees hoop stress that vitrified ceramic was never meant to carry. The formula below sizes the candle bank for a target clarified flow.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| Q | Total clarified flow through the housing | L/h | gal/h |
| J | Specific flux of the candle at reference ΔP | L/(h·m²) | gal/(h·ft²) |
| Ac | Filtration surface area of one candle | m² | ft² |
| n | Number of candles in the housing | — | — |
| ΔP | Operating differential pressure across candle wall | bar | psi |
| ΔPref | Reference ΔP at which J was measured (typically 1 bar) | bar | psi |
Worked Example: Stoneware Filter in a soy sauce clarifier line
Sizing a stoneware candle filter to polish raw moromi pressate at a Yamasa shoyu fermentation plant in Choshi, Japan. The line delivers 1800 L/h of dark soy sauce at 35 °C carrying roughly 200 mg/L of fine koji mould fragments and yeast cells, and the next stage is a pasteuriser that fouls within 6 hours if particulate carry-over exceeds 20 mg/L. Target candle pore rating is 1 µm, candles are 750 mm long × 60 mm OD with a published specific flux of 180 L/(h·m²) at 1 bar reference ΔP, and the available process pressure gives 1.2 bar across the candle bank.
Given
- Qtarget = 1800 L/h
- J = 180 L/(h·m²)
- Lc = 0.750 m
- Dc = 0.060 m
- ΔP = 1.2 bar
- ΔPref = 1.0 bar
Solution
Step 1 — compute the filtration area of a single candle. The candle is a cylinder, so:
Step 2 — compute the per-candle flow at the actual operating ΔP of 1.2 bar (nominal operating point):
Step 3 — solve for the number of candles needed to hit 1800 L/h:
Round up to a standard 60-candle housing. At this nominal point, candles run a clean 4-6 hour cycle between back-flushes and total replacement interval lands around 18 months on Yamasa's solids loading.
Now check the operating-range envelope. At the low end — say the upstream pump drops to 0.6 bar ΔP during a press changeover — per-candle flow halves to about 15.3 L/h, total flow falls to 920 L/h, and the pasteuriser starves. Operators see fill-tank level dropping and assume a clogged filter, but it's just under-pressure. At the high end, if someone throttles the clean-side valve and ΔP climbs to 2.5 bar to chase flow, per-candle flow jumps to 63.6 L/h but you're now running ceramic walls past their fatigue limit — expect candle cracking inside 2-3 weeks, with the first symptom being soy sauce showing visible turbidity in the clean-side sight glass.
Result
The bank needs 60 stoneware candles in a single housing to deliver 1800 L/h of polished soy sauce at 1. 2 bar ΔP. In practice that means a housing roughly 600 mm in diameter and 1.1 m tall, sitting between the press and the pasteuriser, cycling through a 30-second back-flush every 5 hours. The range matters: at the 0.6 bar low end the same bank only delivers 920 L/h and the pasteuriser starves, while at a forced 2.5 bar high end you'll see candle fracture within weeks — the sweet spot is the 1.0-1.5 bar window. If your measured flow comes in 20-30% below the predicted 1800 L/h, check three things in order: (1) tube-sheet gasket bypass — a single missing or rolled EPDM gasket leaks dirty soy back to the clean side and trips the downstream turbidity meter; (2) candle pre-conditioning skipped — new stoneware candles need a 30-minute soak in clean filtrate before first use to wet the pore network, otherwise air locks reduce effective area by 40%; (3) residual koji oil blinding the outer surface, which a 2% citric acid CIP at 50 °C will clear in one cycle.
When to Use a Stoneware Filter and When Not To
Stoneware filters compete with three alternatives in the same particulate band: pleated polypropylene cartridges, sintered metal candles, and pressure-leaf DE (diatomaceous earth) filters. Each wins in a different corner of the operating envelope. Pick on chemistry, flux requirement, and how often you can afford to stop the line.
| Property | Stoneware Filter | Pleated Polypropylene Cartridge | Sintered 316L Metal Candle |
|---|---|---|---|
| Typical flux at 1 bar ΔP | 150-250 L/(h·m²) | 400-800 L/(h·m²) | 200-400 L/(h·m²) |
| Particle retention rating | 0.5-5 µm absolute | 1-50 µm nominal | 1-20 µm absolute |
| Max operating temperature | 180 °C continuous | 60 °C continuous | 400 °C continuous |
| Chemical resistance | Excellent — acids, caustics, solvents | Limited — fails in solvents and strong oxidisers | Excellent except chlorides |
| Service life under fine particulate | 2-5 years with back-flush | 2-8 weeks single-use | 5-10 years |
| Capital cost per m² of area | Medium — —600/m² | Low — €120/m² | High — €2,500/m² |
| Mechanical robustness | Brittle — cracks on impact or thermal shock | Soft — collapses above 3 bar ΔP | Robust — handles 10 bar and water hammer |
| Best application fit | Aggressive chemistry, batch process polish | Cheap, high-flow, clean-water duty | High temperature gas/steam, long life |
Frequently Asked Questions About Stoneware Filter
Back-flush only removes the surface cake. If ΔP refuses to drop below about 0.8 bar after reverse-flow, you're dealing with deep pore blinding — colloidal iron, organic biofilm, or precipitated calcium has migrated 1-2 mm into the wall and surface back-flush can't reach it.
The fix is a chemical CIP, not more air. For iron and scale, soak the candles in 2-3% citric acid at 50 °C for 60 minutes. For organic blinding, 1% sodium hydroxide at 60 °C for 30 minutes. If chemical clean still leaves ΔP high, the candles are sintered shut from heat exposure or simply at end-of-life — replace them.
Always pick the largest pore rating that meets the downstream requirement. A 3 µm candle at the same ΔP gives roughly 2.5× the flux of a 1 µm candle, runs 3× longer between back-flushes, and costs the same. The only reason to drop to 1 µm is if your downstream stage — typically a 0.2 µm sterilising cartridge or a UV reactor — has a documented foulant carry-over limit that 3 µm cannot meet.
A practical check: run a 24-hour trial with 3 µm candles and measure downstream cartridge life. If sterilising cartridges last more than 2 weeks, stay at 3 µm. If they foul in under a week, step down to 1 µm.
Three causes account for nearly every case beyond the ones already covered in the worked example. First, candle orientation — stoneware candles must hang vertically with the closed end down. Mounted sideways or inverted, gas pockets form inside the bore and block flow from the inside surface.
Second, undersized clean-side piping. The candle bank can deliver, but a 25 mm clean-side outlet on a 60-candle housing chokes the discharge — you need to size clean-side pipe for ≤ 1.5 m/s velocity. Third, dissolved gas breakout: if upstream pressure is below the fluid's vapour pressure at temperature, gas comes out of solution inside the candle bore and blocks flow. Add 0.3-0.5 bar of back-pressure on the clean side and the problem disappears.
You can run continuous, but you need a duplex housing — two banks with automated valves so one is filtering while the other back-flushes. A single housing can sustain continuous duty only if your solids loading is below about 50 mg/L and your acceptable ΔP swing is wide (say 0.3 to 2.0 bar over an 8-hour shift).
Above 200 mg/L solids, plan on duplex with a 6-8 hour rotation. Above 500 mg/L, the candle filter is the wrong technology — you want a self-cleaning DE leaf filter or a centrifugal separator upstream to drop the bulk solids first.
Neck cracking is almost never a pressure problem at 1.5 bar — vitrified stoneware handles 4 bar comfortably. It's a torque or thermal-shock problem.
Check the candle nut torque first. Most suppliers spec 20-25 Nm on the top retainer; operators with a long wrench routinely apply 50+ Nm and crush the ceramic neck on the next thermal cycle. Second, look at your CIP sequence — if you switch from 80 °C caustic to 15 °C rinse water in under 60 seconds, the wall sees a 65 °C/min ramp and cracks along the firing grain. Add a 40 °C intermediate rinse and crack rate drops to near zero.
Replace. The economics never work for refurbishment in industrial duty. Once a candle has been through 6-12 months of service, the pore network carries embedded fines that no chemical clean fully removes, and the vitrified body has accumulated micro-cracks from thermal cycling that re-firing won't heal — it just propagates them.
The exception is heritage Berkefeld-style point-of-use units where individual candles cost €40-80 and a careful boil-and-brush gets another season out of them. In a 60-candle industrial housing, the labour to handle and test each candle exceeds the cost of new candles by a wide margin.
References & Further Reading
- Wikipedia contributors. Ceramic filter. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
