Four generations of flotation, and the fourth one is ours
Flotation was invented in the 1970s and has been rebuilt three times since: dissolved air, shallow-layer, and the compact swirl units of the 2000s. Each generation attacked the same problem — bubble size — and each stopped at a different floor. The fourth starts from a ceramic membrane with 20 nm pores.
The chain, in the order the water sees it
Gas passes a 20 nm silicon carbide membrane; the resulting bubbles are cut again in a swirl field; they are injected in stages along the flow path; and the separation zone runs at near-zero velocity so the float does not break up. Four steps, four separate mechanisms, each with a number attached.
SiC membrane gas generation
Gas is forced through a silicon carbide ceramic membrane with a uniform 20 nm pore structure. The result is 40 nm – 50 µm bubbles with a narrow coefficient of variation — the size is set by the pore, not by a pump's whim. No compressor is needed on the liquid side.
Swirl cutting
Coarse bubbles from the membrane are cut again in a rotating field. Operating pressure is only 0.1–0.5 bar, which is why specific energy lands at 0.1–0.3 kWh/m³ instead of the 0.4–0.6 kWh/m³ typical of pressurised dissolved-air systems.
Staged injection
Three to five injection points along the flow path, each releasing a further bubble population. Contact time between gas, liquid and floc is extended by the length of the basin rather than by a bigger tank. Measured TSS removal rises from about 85% with a single injection to 94%.
Zero-velocity separation
The separation zone is designed so horizontal velocity approaches zero. The float layer stays coherent, carries less water, and sludge volume falls by 30–50% against a conventional unit. Effluent turbidity at Cui Jiagou mine water: 0.18–0.20 NTU, against a 0.5 NTU reuse limit.
ARFU-800 series, as measured
| Parameter | 800-II | 800-III | 800-IV | Note |
|---|---|---|---|---|
| Capacity | 25 m³/h | 50–75 m³/h | 100 m³/h | Parallel units for higher flow |
| Specific energy | 0.1–0.3 kWh/m³ (measured 0.18 at Cui Jiagou) | Conventional DAF: 0.4–0.6 kWh/m³ | ||
| Bubble size | 40 nm – 50 µm | Set by 20 nm membrane pore plus swirl cut | ||
| TSS removal | 94% | Conventional DAF typically 85%, site dependent | ||
| Effluent turbidity | 0.18–0.5 NTU | Depends on feed quality | ||
| Membrane | 1200 × 40 mm tubular, 20 nm pore | Silicon carbide, pH 1–14 | ||
| Operating pressure | 0.1–0.5 bar | No high-pressure dissolution tank | ||
| Footprint | ≈50% of an equivalent DAF | Retrofit into existing basins is possible | ||
What the table does not say
- 94% TSS removal is a measured figure on mine water with a specific particle size distribution. On emulsified oil with high surfactant load, the number is lower and we will not quote 94% until we have tested your sample.
- The 0.18 kWh/m³ figure is the whole unit at its design duty, not the generator alone. Adding a recirculation pump or a chemical dosing skid changes the total.
- Retrofit savings depend on what the existing basin already achieves. Replacing a well-run DAF with 90% TSS removal buys you much less than replacing a tired one at 70%.
The energy calculation, done in the open
Specific energy is the only figure that lets you compare two technologies honestly. It is arithmetic, not a claim, and you can reproduce it from nameplate data.
P = Q × Δp ÷ η · E = P ÷ Q
Q = flow (m³/s) · Δp = pressure differential (Pa) · η = pump and drive efficiency · E = specific energy (kWh/m³)
Worked example — generator
12.5 m³/h at 94.0 kPa with η = 0.65 gives P ≈ 502 W and therefore 0.040 kWh/m³. The same architecture in a three-stage configuration reaches 0.030 kWh/m³ because each stage sees a lower differential.
Worked example — deep water
An FRY unit at 70 m³/h and 4.0 kW gives 0.057 kWh/m³. The extra energy is not inefficiency: 30 m of static head has to be overcome to put oxygen at the bottom of a reservoir. A supplier quoting the same kWh/m³ for shallow and deep duty is either guessing or hiding the head.
Why "J per 100 million bubbles" cannot be compared: the denominator is a count, not a mass of gas transferred, so two suppliers can produce the same number with completely different oxygen delivery. Likewise, a passive static device and a gas-injection unit cannot be subtracted from one another — their energy bases differ. We explain that trap on the limits page.
Two more separation technologies behind the generators
Hypergravity swirl membrane filtration
A rotating field substitutes centrifugal force for gravity, raising the separation driving force by an order of magnitude — about 2000 g — and shrinking the equipment accordingly. The same rotation scours the membrane surface, which is what keeps flux stable on oily feed. Trade-offs: rotating equipment needs maintenance, and feed solids have a ceiling. Detail →
Silicon carbide ceramic membranes
SiC survives pH 1–14, so strong acid and caustic cleaning remain available when flux drops — the regime polymeric membranes cannot take. Pure-water flux ≥2000 LMH/bar, contact angle 0.3°, hardness 9.2 Mohs, 20-year material life quoted. Detail →
Reference material
ISO 20480-1:2017
Fine bubble technology — general principles, Part 1: terminology. The definitional basis for "fine bubble", "microbubble" and "ultrafine bubble" used throughout this site.
ISO 20480-3:2021
Part 3: generation methods. The classification we use when we describe how a generator works — pressurised dissolution, shear or swirl, porous membrane diffusion, electrolysis.
Status: both published and current. The full 52-item picture, including Chinese, Japanese and Korean national standards and the four areas where no standard exists, is on the standards page.
Want this applied to your duty point?
Send the flow, the pressure available and the target parameter. We will return a model, a specific energy figure at that point, and the assumption that would break it.
Talk to an engineer → All 39 models