Ultra-Heavy Oil Defeats Gravity Separation: The Sequencing Fix Behind 150,000 to 40 mg/L

Ultra-heavy oil is the stream that makes conventional oil-water equipment look foolish. At a Ningbo electric-desalting wastewater project, SINOKLE was asked to bring 150000mg/L at the inlet down to 40mg/L at the outlet. Two units did it, a coalescing pressure deoiler and a CDFU dissolved-air cyclonic flotation unit, and the interesting part is not either machine on its own but the order they sit in.

Anyone who has specified industrial wastewater equipment knows the rule: selection sets the ceiling of a process, and sequencing sets the floor it will actually operate at. This project is a clean illustration of both halves.

The density problem nobody can engineer around

Pick up a datasheet for almost any standard coalescing deoiler and the separation step is the same. Droplets grow on the packing, then rise because oil is lighter than water. That assumption collapses with ultra-heavy oil. With API specific gravity below 10, the density gap to water is typically within 0.02g/cm3. Buoyancy is proportional to that gap, so the driving force for gravity rise is essentially zero: the droplets grow and then simply sit there.

Which is why a machine bought off a catalogue will not hold up here. The coalescing section has to be rebuilt around the fluid, not the other way round.

Three adjustments that make coalescence work at near-zero density difference

• Packing selection. The medium needs high affinity for ultra-heavy oil specifically, and it needs to survive contact with it. At ambient temperature this oil is extremely viscous, and once the packing surface scales it is hard to clean. Selection is a deliberate balance struck between oleophilicity and fouling resistance, and it has to account for temperature resistance at the same time.
• Flow-channel design. The coalescing section gets more effective contact area and longer residence in the packing bed, raising the odds that a tiny droplet meets another one. Channel cross-sections are reshaped in parallel to prevent the local blockage and maldistribution that high viscosity causes.
• Operating pressure. Within a sensible range, raising system pressure increases the driving force pushing the oil-water mixture through the coalescing bed, so density difference stops being the only variable deciding whether separation happens at all. Pressure-mode operation replacing gravity-mode operation is the core feature distinguishing this equipment from a traditional coalescer: the driving force shifts from leaving it to nature to doing active work.

Together those three answer one question. Under the extreme condition where the oil-water density difference is nearly zero, how do you still make tiny droplets collide, coalesce and grow efficiently, and then get them out through a designed flow path?

What is actually left in the water afterwards

Once the front end has removed the bulk of the 150000mg/L, what remains in the water is mainly dispersed and micro-emulsified oil, running somewhere in the hundreds to thousands of mg/L. The treatment goal changes character at this point. It is no longer bulk oil removal, it is precise control.

Inside CDFU: four mechanisms running at once

CDFU, or Cyclonic Dissolved Gas Flotation Unit, is an honest name, because the unit genuinely fuses several technical routes and runs four separation mechanisms simultaneously inside a closed pressure vessel.

• Cyclonic centrifugal separation. Oily water enters the tank tangentially and forms a high-speed cyclone. In that centrifugal field the denser water is flung to the outer wall while the lighter oil and bubbles gather in the central low-pressure zone. This step handles the primary spatial distribution, driving oil and water into their own zones first.
• Ultra-fine bubble generation. The integrated generator produces ultra-fine bubbles with diameters below 30μm. That figure is not arbitrary. If bubbles are too large, collision probability and attachment efficiency with tiny droplets are poor; if they are too small, buoyancy is insufficient to lift what they catch. 30μm is the optimal range found after extensive operating-condition testing.
• Dissolved-air flotation. Part of the effluent is pressurised, saturated with air and returned to the tank. When pressure drops sharply, dissolved gas is released as micro-bubbles. These bubbles are extremely numerous and evenly distributed, and they are the main force behind deep oil removal.
• Coalescence demulsification. Coalescing demulsification packing sits in a specific zone inside the tank and performs a secondary demulsification and coalescence pass on incompletely separated emulsified droplets. This is the step that catches the holdouts, the droplets whose interfacial films are especially stubborn.

These four are not a simple one-plus-one-plus-one-plus-one pile-up. They run in a series with timing, division of labour and handover: cyclonic separation does the coarse split, ultra-fine bubbles and dissolved-air flotation carry the main capture, and coalescence demulsification does the final cleanup.

Why the deoiler comes first, and what happens if you swap them

Any process engineer answers this one from load logic. Put CDFU in the first stage and 150000mg/L instantly overloads the micro-bubble generation system; no matter how many micro-bubbles the dissolved-air release produces, they cannot cope when oil outweighs water. On top of that, the high viscosity of ultra-heavy oil creates huge shear resistance in the cyclone section, pulling cyclonic separation efficiency down sharply.

The coalescing pressure deoiler is inherently suited to the opposite conditions. Its separation mechanism does not depend on the capture precision of tiny bubbles; it depends on the droplets colliding, coalescing and growing on their own. So the front stage uses coalescence to shave the peak, bringing concentration down from 150000mg/L into a range that flotation can treat economically, and the rear stage uses CDFU for fine finishing, polishing a few hundred mg/L down below 40mg/L. The load-distribution logic between the two is clean, each performs its own role, and that is what lets the train run stably for years rather than for a commissioning week.

The technical content of the equipment itself is one thing. Placing each piece at the correct position in the process chain and letting it do what it is best at is the real craft of engineering implementation.