Refineries are chasing “polish” water: GAC vs. AOPs—and why RO wins for reuse

Refinery wastewater is a cocktail of oil, high-strength organics and toxics—COD (chemical oxygen demand) often in the hundreds to thousands of mg/L; ammonia; phenols; BTEX; sulfides; heavy metals (climate-policy-watcher.org). Catalytic-cracking streams alone can range 500–2800 mg/L COD with oil & grease up to 2600 mg/L (climate-policy-watcher.org).

Standards force a final cleanup. India’s CPCB (a widely cited benchmark alongside rules in Indonesia/ EU/ US) holds petroleum refinery effluent to COD ≤125 mg/L, BOD (biochemical oxygen demand) ≤15 mg/L, oil & grease ≤5 mg/L, phenol <0.35 mg/L (corpseed.com). Even after biology, effluent typically carries tens of mg/L COD and residual hydrocarbons—a pilot still saw ~ppm-level total hydrocarbons in treated effluent (researchgate.net).

Reuse is tougher. Cooling or process makeup often means <10–15 mg/L COD, ultralow turbidity and tight ionic control. In one case study, adding reverse osmosis (RO) polishing delivered COD below 45 mg/L, hardness <8 ppm, Fe ~0.04 ppm and ammonia <1.1 ppm—with about 90% reuse of treated wastewater and a 20–25 tonne/hour cut in fresh intake (researchgate.net).

Effluent targets and residual loads

Refineries confront complex influent—high COD, oil & grease, and trace toxics—before they ever get to polishing (climate-policy-watcher.org). Discharge norms (e.g., CPCB COD ≤125 mg/L; BOD ≤15 mg/L; oil & grease ≤5 mg/L; phenol <0.35 mg/L) push a final barrier after biological treatment (corpseed.com). Even then, studies report residual COD in the tens of mg/L and ppm-level hydrocarbons in treated effluent (researchgate.net). For reuse, targets tighten to <10–15 mg/L COD and near-ion-free water—where RO steps in, with one pilot cutting fresh intake by 20–25 t/h and reusing ~90% of flow (researchgate.net).

GAC adsorption as a polishing barrier

Granular activated carbon (GAC) relies on extreme porosity (specific surface area often BET >1000 m²/g) to adsorb hydrophobic trace organics like residual oils, phenols and BTEX. Refineries commonly deploy activated carbon contactors after biology to strip these molecules at low concentrations.

Results are strong but capacity-limited. In refinery pilots, biologically activated carbon (BAC—GAC colonized with biofilm) posted ~65% average total organic carbon (TOC) removal over long runs; a GAC column saturated in ~28 days, while BAC sustained organics removal for 84 days (researchgate.net). In the same work, pairing AOP with BAC drove effluent TOC to ~4–8.5 mg/L; notably, the standalone AOP step had only lowered TOC by 10–18% (researchgate.net).

Operating realities: beds need backwashing; media is regenerated or replaced (often every few months or less, load-dependent). Used GAC prices are often a few US dollars per kg; in industrial practice, total GAC operating costs may run roughly $1–5 per m³ treated, with low energy use but sizable footprint and consumables logistics. GAC does not remove salts, ammonia or metals unless specially formulated (researchgate.net).

AOPs for refractory organics oxidation

Advanced oxidation processes (AOPs)—ozone (O₃) and/or hydrogen peroxide (H₂O₂) plus UV or catalysts—generate hydroxyl radicals (·OH) that non‑selectively attack persistent organics. Plants often pair UV reactors with UV disinfection units to deliver the radiation dose, and accurate peroxide feeds via a dosing pump when H₂O₂ is used.

Removal rates of targets can be striking. One lab H₂O₂/UV/O₃ setup degraded ~93.8% phenol at 1000 ppm H₂O₂; other studies cite 90–99% removal of aromatic hydrocarbons or MTBE under optimized conditions (mdpi.com; mdpi.com; researchgate.net). A combined UV/H₂O₂/O₃ treatment on real petrochemical effluent achieved 94% COD and 68% TOC removal in 30 minutes (researchgate.net).

The catch: partial mineralization is common. AOPs frequently convert large molecules to small acids or aldehydes, leaving residual COD; in one pilot, O₃/UV or H₂O₂/UV trimmed TOC by only ~10–18%, and BAC did the heavy lifting afterward (researchgate.net). A multistage refinery flow (Yang et al., 2021) only reached >94% overall COD removal when biological steps were combined with an O₃/GAC polish; a final O₃/activated-carbon tank pulled COD to ~50 mg/L and TOC to low double-digits (nature.com; nature.com).

Energy and chemical intensity is significant. Effective control of COD or micropollutants often runs >100 mg/L O₃ or H₂O₂ doses and high UV exposure (“many dozens of kWh”) (mdpi.com; researchgate.net). One test needed 3060 mg/L H₂O₂ at pH 9 to hit 98% phenol removal with H₂O₂/UV (mdpi.com). Ozone generators use ~10–20 kWh per kg O₃; UV lamps can consume ~1–2 kWh/m³ per 1000 mJ/cm² UV-dose. A refinery sour-water techno-economic study put an RO+UV/H₂O₂ train at ~1.2–2.5 kWh/m³ with energy recovery (pubs.rsc.org)—and AOP steps on their own add a similar order of magnitude.

Byproducts are manageable but present. Ozone commonly oxidizes aromatics to short-chain acids, which can lower pH; in the integrated O₃/AC study, volatile fatty acids fell to ~3.3 mg/L, essentially benign (nature.com). AOPs do not remove inorganics (salts, metals, ammonia), so downstream polishing is typical (researchgate.net).

Membrane filtration for reuse-grade water

When the goal shifts from discharge to reuse, RO is the workhorse: typical industrial elements reject >99% mono- and divalent ions and remove suspended solids/colloids. Vendors bundle RO, nanofiltration (NF) and ultrafiltration (UF) in integrated membrane systems for industrial water.

On refinery secondary effluent, pilots report >98% rejection of phenolic compounds and >90% water recovery, with permeate conductivity <100 μS/cm, hardness <8 mg/L, COD <45 mg/L, metals ~0.04 mg/L and ammonia <1.1 ppm—well under reuse thresholds (pubs.rsc.org; researchgate.net). At Genoa, stable permeate quality over days translated to 20–25 t/h of fresh-water savings (researchgate.net).

Pretreatment is non-negotiable. UF routinely protects RO from oil and solids; one oil‑tolerant UF achieved >70% removal of oil and TSS, producing permeate <10 mg/L oil and <30 mg/L solids (polycera.com). Many refineries stage UF before RO via an ultrafiltration module, and in some cases add nanofiltration as intermediate polishing.

Operating windows vary by salinity. Typical brackish setups run ~15–25 bar with 10–30 L/m²·h flux—standard duty for a brackish-water RO. In very salty sour-water pilots, a two-pass seawater RO membrane held ~90% recovery at ~20 bar with 10–15 L/m²·h flux, consistent with seawater RO designs (researchgate.net).

Energy and economics have improved. Without energy recovery, RO on high‑TDS streams draws ~3–5 kWh/m³; modern devices push this to ~1–2 kWh/m³ (pubs.rsc.org; pmc.ncbi.nlm.nih.gov). In one refinery case, energy landed at 1.2 kWh/m³ with recovery and operating cost at ~€0.41/m³ permeate, with ~375,000 m³/yr potable water saved and 417,000 m³/yr discharge reduced—about €286,000/yr and a 4‑year payback (pubs.rsc.org).

Limitations are practical. RO cannot take free oil; feeds must be virtually oil‑free. Fouling (biofilm, scaling) needs periodic cleaning (CIP). RO separates rather than destroys organics, so some low‑molecular‑weight compounds may pass unless designs include two‑pass operation; brine management is required (pubs.rsc.org).

Technology comparison and selection

Trace organics: GAC adsorbs broadly to low concentrations and is ideal when biologically treated effluent holds sub‑100 mg/L residual organics; it often delivers 50–90% removal of what’s left until breakthrough. AOPs oxidize hard‑to‑biodegrade compounds and readily post 90%+ removal for target molecules like phenol or BTEX (mdpi.com; mdpi.com)—but overall TOC often only drops by tens of percent without hefty doses. Field work shows AOP alone may shave ~10–20% off TOC, while adding BAC can push effluent TOC ≈5 mg/L (range ~4–8.5 mg/L) (researchgate.net).

Efficiency vs. cost: AOP’s deep oxidation can come with high energy/chemical bills. Meeting phenol limits with ozone might need >30 mg/L O₃ (tens of kWh/m³) plus GAC polishing; photocatalytic or Fenton configurations carry similar burdens (researchgate.net). GAC costs hinge on media turnover—roughly $1–5/m³ including handling—while energy use is low. For extremely tight limits (e.g., <1 mg/L phenol), AOP + GAC pairings are common.

High‑reuse scenarios: For cooling or process reuse, membranes do what GAC/AOP cannot—remove salts at scale. Case studies show RO meeting or beating reuse specs with >90% recovery and >98% rejection of phenols (researchgate.net; pubs.rsc.org). Many plants adopt UF pretreatment → RO, with optional oxidation or adsorption inter-stages as loads dictate. Trends show growing RO/NF adoption for maximum reuse or near‑zero liquid discharge in water‑scarce regions (pubs.rsc.org).

Key performance and cost markers

GAC/BAC filters have demonstrated ~65% long‑term TOC removal before saturation; standalone AOP pilots reach ~93–99% removal of model phenols or BTEX but often leave >20% of initial TOC; membrane pilots show >90% RO recovery and >98% rejection of organics (phenols), yielding permeate with <50 mg/L COD. One RO case reduced fresh intake by ~20–25 t/h and saved ~€286k/year with a 4‑year payback—alongside ~375,000 m³/yr potable water saved and 417,000 m³/yr discharge cuts (researchgate.net; mdpi.com; pubs.rsc.org; researchgate.net).

Sources and further reading

Analysis draws on case studies and reviews of refinery wastewaters covering performance, efficiencies and recoveries: mdpi.com; researchgate.net; pubs.rsc.org; researchgate.net; corpseed.com. Baseline wastewater characteristics and catalytic‑cracking ranges: climate-policy-watcher.org. Ozone/AC integration and volatile acids: nature.com and nature.com. UF pretreatment performance: polycera.com. RO energy benchmarks: pmc.ncbi.nlm.nih.gov.