Polymer Flooding Overloads Offshore Separation: Two CDFU Stages on a Bohai Platform

A platform in the eastern Bohai Sea had already tried the obvious levers. More demulsifier, longer residence time in the jet-pump power-fluid separator, tighter operator attention. None of it held. What finally moved the effluent was a change of equipment class: two stages of CDFU cyclonic dissolved air flotation inserted after the three-phase separator, with the recovered oil routed back upstream.

The case is worth reading closely because it is not a story about a bigger vessel. It is a story about what happens to a separation train when the chemistry of the feed changes underneath it.

The reservoir set the stage, the flooding method spoiled it

Exploration on this structure began in 1996, the first well was drilled in 2000, and system production assessment closed in 2005. Proven petroleum in place is 5222.07x10^4 m3, of which heavy viscous oil is 4187.87x10^4 m3 — 85% of the total. Heavy oil at that share makes recovery method a strategic question rather than an operating preference, so steam huff-and-puff and chemical flooding were pushed into service early.

Both work. Both also leave a feed that behaves nothing like the fluid the original separator train was sized for. Long-term polymer injection changes the rheology and the interfacial behaviour of the produced fluid in ways that show up as four separate failures at once:

• Emulsification intensifies and the separation load climbs. Polymer chains stabilise the oil-water interface to a degree that ordinary gravity settling cannot break. Oil in the separated water rises, water in the export crude rises with it, and neither stays inside target.
• Chemical response falls away. Demulsifiers that performed well on conventional feed lose a large part of their effect. Sites compensate by dosing harder, which converts a chemistry problem into a permanent line item in the operating budget.
• Heat exchangers foul and coke. Polymer-laden fluid deposits on hot surfaces, so units come offline for cleaning more often, and every cleaning window is lost production.
• Oily sludge accumulates. The sticky byproduct is difficult to handle on a deck with limited space and it keeps disturbing downstream facilities.

The feed arriving at the separator

On this platform the power-fluid separator was receiving produced fluid at 35-60 C with oil content up to 20%. The stream carried large amounts of foam and was severely emulsified, so the water leaving it was poor across the board. Downstream processes were being asked to compensate for an upstream that could not deliver, and the whole production system was running below its design envelope as a result.

That framing matters for selection. If the problem were simply insufficient retention time, a larger vessel would work. Here the emulsion was stable enough that time alone no longer resolved it, which pointed toward a unit that could apply mechanical energy and bubble surface area together.

Why cyclonic dissolved air flotation, and why two stages

The project team compared the mainstream options on applicability, maturity and demonstrated performance, then chose CDFU, the cyclonic dissolved gas flotation unit, on the strength of operating records rather than pilot data alone. Comparable duty had already been run at PetroChina Dushanzi Petrochemical on electro-desalting wastewater, at Tuha oilfield on oily wastewater, and at a CNOOC Bohai offshore platform on oily water.

The mechanism is a deliberate marriage of two effects that are usually installed separately. Cyclonic centrifugal separation provides a rotational field that drives oil droplets and bubbles toward the axis and raises the probability that they meet. Dissolved gas flotation supplies the bubble inventory that makes the meeting useful. Water saturated with gas passes through a large-diameter release device that produces micro-nano bubbles in the 5~30 μm range without clogging internally; the stream carrying those bubbles enters the vessel tangentially from the bottom, so the weak cyclonic field and the bubble swarm act on the same volume at the same time.

Gas dissolution itself is handled by German EDUR gas-liquid two-phase pumps paired with a high-precision gas flow controller, which the pump maker rates at 100% saturation. Saturation is the number that decides how much bubble surface area is available per cubic metre of feed, and it is the first thing to verify when a competing skid claims similar performance.

Two stages exist for a reason that has nothing to do with redundancy. The first stage takes the bulk of the free and mechanically dispersed oil; the second works on what survives, which is largely the emulsified fraction. Emulsified droplets captured by ultra-micro bubbles aggregate, coalesce and demulsify under cyclonic action, forming clusters heavy enough to separate quickly. That demulsified oil is collected and returned to the upstream three-phase separator for a second pass rather than being written off as slop.

What fifteen commissioning samples showed

The operating test ran fifteen samples across the inlet and outlet of the new unit. Inlet oil content sat in a narrow band between 1187.2 and 1365.5 mg/L, which is the signature of a steady feed rather than a series of upset conditions. Outlet values ran from 106.8 to 122.5 mg/L. Single-sample separation efficiency worked out between 90.26% and 92.02%.

Across the whole set, average oil content fell from about 1275.2 mg/L at the inlet to about 114.7 mg/L at the outlet, an average reduction of 1160.5 mg/L and an oil removal efficiency of 90.99%. The design requirement was 80%. The margin is not a rounding detail: it is the difference between a unit that meets specification on a good day and one that holds specification when the feed drifts, and it is what allows the downstream facilities to be run at their intended load instead of being flogged to compensate.

What the gain actually delivered

Three consequences followed, and only the first is about water quality.

• The separator no longer sets the ceiling on platform throughput, so production can be scheduled on reservoir and market logic.
• Downstream treatment facilities receive a much lighter load, which shows up as lower chemical consumption and fewer backwash cycles rather than as a single dramatic saving.
• Recovered oil goes back into the process stream instead of into the sludge pit, which reduces both disposal cost and the deck space tied up in byproduct handling.

Where the same skid would not help

Cyclonic dissolved air flotation is a physical process. It removes free oil, dispersed oil and emulsified oil that mechanical shear and bubble attachment can destabilise. It does not remove dissolved hydrocarbons, and it does not touch dissolved salts, so it cannot substitute for a de-oiling step ahead of a membrane train when the target is dissolved organics.

Two operating conditions also deserve scrutiny before committing deck space. First, the unit needs the feed to arrive as a liquid stream of predictable flow; severe slugging forces the level control to fight the cyclone, and efficiency drops before any alarm sounds. Second, the recovered oil stream has to have somewhere to go. On a platform where the three-phase separator is already at its water limit, the recycle path is the constraint that decides whether the second stage earns its footprint.

For polymer-flooded offshore fields the conclusion is fairly direct. When the emulsion has become mechanically stable, adding residence time buys very little and adding a cyclonic flotation stage buys a great deal — provided the stage is placed after the three-phase separator, sized for the full produced-fluid rate, and given a return path for the oil it recovers.