Coking Acidic Water: Four Traits That Break Conventional Oil Removal
A coking plant does not produce one wastewater. It produces a family of them, and they reach the treatment plant from different corners of the battery: moisture that entered with the coal blend, water released during the coking reaction, distilled ammonia liquor, and gas condensate drawn out of the gas train. Operators sort the result into excess ammonia liquor, phenol-cyanide wastewater, dust-removal wastewater, gas seal water and desulfurization waste liquor. Treating the mixture as a single ordinary effluent is where many designs go wrong, because it is complex and variable, the volumes are large, and the pollutant load is high. Oil content on its own can reach 13.3%-16%, COD runs far above anything a municipal plant would see, and ammonia nitrogen, sulfides and volatile phenols all arrive in the same bucket.
The four properties that actually drive selection
Strip away the labels and the difficulty reduces to four measurable properties, and every one of them removes an option from the catalogue.
- Coke powder. Suspended solids load is high and abrasive, so any internal part with a narrow passage becomes a wear item and a blockage risk.
- Refractory organics. Phenols and cyanides dominate the dissolved fraction, and the biodegradability ratio is correspondingly poor. Sending the raw stream straight to biology does not work.
- Emulsified oil. The oil is not free product sitting on top of the water. It is held in a stable oil-water mixture that ignores a plain retention tank.
- Moving quality. A coking push, a coal blend change or a maintenance event can shift oil, solids and phenols at the same time, so the plant must absorb shocks rather than operate at a single design point.
Why chemical demulsification is a poor first move
The reflex answer is to dose a demulsifier and let the oil float. It works, and then it hands you a second problem: the separated oil, the floc and the residual chemical leave the plant as oily sludge, which is hazardous waste with a disposal cost attached to every cubic metre treated. At coking-plant scale that line item grows quietly and never shrinks. A physical route avoids generating the waste in the first place, keeps the recovered oil clean enough to return to the oil circuit, and does not add a reagent whose dosage has to be chased every time the influent shifts.
The train, stage by stage
The configuration SINOKLE builds places a coarse filter first, CDFU cyclonic dissolved-air flotation at the working core, and a KHC coalescing unit in the polish position. The coarse filter looks unglamorous but is load-bearing: it removes large suspended particles and shields everything downstream from abrasion and blockage.
CDFU brings centrifugal separation, dissolved-air flotation and ultrafine bubble generation into one vessel. Oil removal exceeds 90% and suspended solids removal exceeds 80% inside a residence time of 5 minutes or less. The mechanism behind those figures is bubble size. CDFU generates microbubbles in the 5-30 μm range, while conventional flotation works with bubbles of 300-1,000 μm. Smaller bubbles deliver far more interfacial area per unit of gas and a much higher chance of attaching to a fine oil droplet — precisely the fraction that gravity settling cannot lift.
Polishing is handled by a KHC high-efficiency coalescing de-oiler whose fibre media filters at 0.1 μm precision. Remaining fine droplets are forced to meet, merge and rise as larger bodies, and effluent oil content is driven below 10 ppm, the level at which the water becomes usable rather than merely disposable.
What the Luoyang retrofit measured
At a petrochemical company in Luoyang, on a heavily emulsified acidic water stream, the recorded performance was influent oil at 2,000 ppm, effluent below 50 ppm, and suspended solids removal above 98%. Effluent ran clear and transparent, with oil content and suspended solids both holding below 20 mg/L, which met the reuse standard the site was targeting. The figure worth pausing on is not the peak removal but the stability: the unit held those numbers through fluctuations of roughly 50% in both water quantity and quality. That is what separates a working plant from a demonstration.
Footprint, enclosure and the operating bill
Three engineering choices show up on the balance sheet. CDFU is compact, occupying around 10% of the footprint of a conventional flotation cell, and a single unit can carry up to 500 m³/h, so land take and civil works shrink with it. The train then runs fully enclosed and pressurised, which removes the hydrogen sulfide and oil vapour leakage path that makes coking water plants unpleasant and, in the wrong weather, dangerous — the design intent is intrinsic safety rather than ventilation and monitoring. Modular skid construction suits retrofits: the construction schedule shortens by more than 50%, and the control system supports remote monitoring and unattended operation, which is where most of the O&M saving actually comes from.
The environmental arithmetic follows. No chemical addition means no secondary pollution. Physical separation recovers waste oil at a rate of 99%. Enclosed operation means no exhaust gas escape, and the recovered oil is a product rather than a disposal problem. A chemical company in Henan Province running highly emulsified acidic water through the same train produced clear, transparent effluent, which is clean production in the literal sense rather than the brochure sense.
Where this design stops being the answer
Three boundaries are worth stating plainly. The train removes oil and solids; it does not fix a biological problem, so where the discharge route requires ammonia and phenol destruction a biological stage still follows and its nutrient balance must be managed separately. Second, the coarse filter has to stay in service — bypass it during a maintenance window and the CDFU and KHC internals pay the bill. Third, the numbers above come from one influent. A stream with materially different oil speciation, or a much heavier solids load, needs its own pilot rather than a copied datasheet.