To hit phenol limits of <0.5–1 mg/L and push reuse above 60%, refineries are layering granular activated carbon, advanced oxidation, and reverse osmosis. Each option hits the load differently — and the cost curve even harder.
A major Chinese petrochemical complex now treats about 67,000 m³/day of effluent and reuses roughly 70% as cooling‑tower makeup, according to Suez. That’s the direction of travel as regulations tighten — China’s petrochemical standard pegs effluent COD (chemical oxygen demand; a proxy for organic load) at 60 mg/L (MDPI) — and as “refractory” organics (phenols, PAHs/polycyclic aromatic hydrocarbons, pesticides) slip past primary/secondary treatment.
Closing that last gap takes tertiary polishing that removes >90% of the remaining COD/BOD (biochemical oxygen demand) and virtually all phenols. In practice, integrated systems — biological, coagulation, adsorption, advanced oxidation — target >95% total organics removal. The buildouts increasingly include ultrafiltration (UF) and reverse osmosis (RO), with polishing steps such as granular activated carbon (GAC) and advanced oxidation processes (AOPs).
Trace‑organics polishing with granular activated carbon
GAC (granular activated carbon) is the mature adsorption workhorse for tertiary “polish.” It has high affinity for hydrophobic and aromatic compounds, including phenols and cresols, and can drive very high removal at sufficient contact time. Pilot studies report COD removals of ~95–97% (MDPI). One packed GAC column (80% bed packing) achieved 96.7% COD removal from refinery wastewater at pH≈5.7 (MDPI).
In an integrated three‑step flow — electrocoagulation + bioreactor + GAC — the final GAC column (agricultural‑waste carbon) removed ~97% of remaining COD and essentially 100% of phenol and cresol (MDPI). Bench tests with wood‑based GAC yielded ~95% BOD and 88% TOC (total organic carbon) removal (MDPI). These figures imply GAC can typically push COD from a few hundred mg/L down to ≪50 mg/L. In practice GAC often provides the final, high‑purity polish after other treatment steps (MBR, AOP, etc.).
Trade‑offs are clear. GAC doesn’t destroy pollutants; it accumulates them, and media eventually saturate. Spent carbon must be regenerated or replaced, often via multiple columns so regeneration can occur on‑line. One cost analysis found ~70% of total treatment OPEX went to maintenance including GAC regeneration (MDPI). On the upside, GAC beds use minimal energy (pumps/fans), impose a moderate footprint, generate little waste (spent carbon) and no toxic by‑products; on the downside, high suspended solids or oil can foul carbon quickly and adsorption can falter on highly soluble or very low‑molecular‑weight organics. Operators commonly specify granular media such as activated carbon at this polishing step.
Advanced oxidation for refractory loads
AOPs (advanced oxidation processes — ozone, UV/H₂O₂, Fenton chemistry, photocatalysis, persulfate) generate hydroxyl radicals (•OH) to chemically break down organics and are typically deployed as tertiary or pre‑RO steps. Bench and pilot data show high conversion when optimized: a combined electrocoagulation + ZnO/photocatalytic system removed ~94% of COD from a ~900 mg/L refinery feed within 60 minutes (MDPI).
Fenton oxidation (H₂O₂ + Fe²⁺) has achieved >95% COD reduction when followed by polishing; one upgrade added Fenton + UF (H₂O₂ dosing ~300 mg/L with H₂O₂:Fe = 3:1, pH ~2.3) and cut COD from ~160 mg/L to <50 mg/L (MDPI). AOPs excel on aromatics like phenols. Combined ozone/photo‑oxidation (“catalytic ozonation”) has been shown to make secondary effluent meet even 60 mg/L COD standards (MDPI; MDPI). One pilot of catalytic ozonation (γ‑Al₂O₃ catalyst, ~40 mg/L O₃, 1 h contact) stabilized secondary petrochemical wastewater over 9 months and achieved ~38% COD removal in one pass, with effluent dropping to ~50 mg/L (MDPI).
The cost side is material. Ozone must be generated on‑site; UV/H₂O₂ needs lamp power and H₂O₂ dosing. In one case, adding UF to a Fenton+AC pretreatment raised removal from ~150 mg/L COD to <50 mg/L (via 3:1 H₂O₂/FeSO₄ at 300 mg/L) but increased OPEX by roughly 20% (~+1 USD/m³) for the added unit and chemicals (MDPI; MDPI; MDPI). Typical AOP OPEX (ozonation or UV) can exceed $0.50–1.00/m³ for reagents/energy, depending on dosage and loads. In a pilot targeting COD<50 mg/L, adding UF post‑Zn/Fe Fenton increased OPEX by ~20% (∼$5–6/m³ total) (MDPI). When UV is selected, tertiary trains often incorporate dedicated UV systems.
Control and synergy matter. AOPs may form by‑products (e.g., bromate from ozonation of bromide‑bearing waters), so downstream contactors or adsorption — often with activated carbon — may still be needed. One municipal effluent study found an “optimal” ozone dose of 0.20–0.30 g O₃ per g DOC combined with ~10 mg/L PAC (powdered activated carbon) maximized trace‑organics removal (ScienceDirect). Footprint is moderate to high CAPEX (ozone generators, UV reactors, piping) but compact in layout, with little solid waste. Flexibility is a draw when specific hard‑to‑treat chemicals (e.g., TCE, nitrosamines, endocrine disruptors) are targeted.
Membrane filtration for high‑grade reuse
For very high reuse targets, membranes physically remove virtually all suspended and dissolved contaminants once upstream polishing tames the organics. RO (reverse osmosis) is the step when the goal is “nearly fresh” water for cooling makeup, process use, or even indirect potable reuse. In one Sino‑foreign refinery project, an integrated UF→RO train produced permeate at about 40 mg/L COD from much higher feed — corresponding to >95% removal of organics — and that water was reused for cooling (65–70% of plant potable needs) under China’s water‑reuse regulations (WaterTech Online). As a rule of thumb, RO can achieve >99% removal of dissolved organics and salts, yielding the highest reuse quality. Pretreatment commonly includes ultrafiltration to protect the RO stage, while the RO itself is typically executed on RO skids sized for brackish‑range TDS.
Designs vary, but the integration logic is consistent: membranes require careful pretreatment. An Indonesian pilot used multimedia filtration and UF ahead of RO; the RO permeate met cooling‑water specs, while the RO concentrate — the ~30–35% not recovered — was treated by an MBR and then GAC polishing (Taylor & Francis). That concentrate train maps directly to technologies like membrane bioreactors followed by activated‑carbon polish.
Performance constraints are known. RO recovery rates in practice are often 60–80%, and high water‑reuse goals push equipment toward the upper end. Fouling risks (silica, biofilm, oil) mean frequent cleaning/backwash of UF is usually needed; in the Indonesian case, UF cleaning every 14 days kept RO flux stable (Taylor & Francis). RO is energy‑intensive: typical industrial membranes use ~3–6 kWh/m³, plus pretreatment and pumping. OPEX runs on the order of USD 0.5–1.0/m³ for power plus membrane chemicals, with about a ~$1/m³ uplift when UF and membranes are added on top of conventional treatment (MDPI). Capital costs are high (high‑pressure pumps, membrane skids) and footprint is moderate. In water‑scarce regions, “zero liquid discharge” schemes even evaporate or crystallize RO reject, but more commonly the concentrate is treated by MBR/GAC to avoid waste brine.
Comparative performance and economics
Effectiveness: both GAC and AOP attack different parts of the residual load and can hit very high removals. GAC alone can remove ~95–97% of remaining COD and all phenolics in tertiary polishing steps (MDPI; MDPI). AOPs can similarly achieve >90% COD removal under ideal conditions (MDPI; MDPI). In practice, many facilities sequence them: an AOP stage (e.g., ozonation) to oxidize non‑adsorbables, followed by GAC to capture oxidized by‑products and any residuals. For trace organics (micropollutants), ozone+carbon combinations often outperform either alone (optimal ozone 0.20–0.30 g O₃/g DOC plus ~10 mg/L PAC; ScienceDirect).
Cost and footprint: GAC has lower energy demand but recurring costs tied to regeneration and media; one analysis attributed ~70% of total treatment costs to maintenance (largely GAC regeneration; MDPI). AOPs incur significant chemical/electric loads (ozone generators, H₂O₂, UV lamps). In one pilot aiming for COD<50 mg/L, adding UF post‑Zn/Fe Fenton increased OPEX by ~20% (∼$5–6/m³ total; MDPI). Membrane reuse is generally the most expensive tier; besides CAPEX, it added about $1/m³ to pretreatment cost in that study (MDPI). The business choice is typically between GAC columns, an O₃/UV reactor train, and a high‑pressure RO system, weighted against the value of reuse water.
Regulatory context and system design
Trends point to more advanced tertiary in Asia and Indonesia. New refinery projects in China mandate ≥65% internal reuse (Suez). Research and regulation emphasize AOPs (Fenton, photocatalysis, ozonation) for hard‑to‑treat effluent, while adsorption — conventional and novel carbon adsorbents — remains a core polish step (MDPI). Pilot and full‑scale results converge on modern trains that use all three tools — biological plus AOP plus membranes plus GAC — for ultra‑clean discharge or high reuse.
The layered approach is pragmatic: GAC adsorption is simple and very effective (often >95% removal; MDPI; MDPI) but yields a spent media stream. AOPs oxidize persistent compounds and reduce toxicity, at higher energy/chemical cost. RO can then deliver permeate around ~40 mg/L COD or lower (WaterTech Online) with recoveries ~65–70% in high‑reuse configurations. Designs should reflect local limits; Indonesia follows stringent standards similar to China’s GB30571 (COD ~50–100 mg/L), so dose, contact time, and surface area are tuned to those numeric targets.
