Refineries Count On Quiet Workhorses: Inside the Long, Shallow Tanks That Make or Break Wastewater Compliance

In refinery wastewater treatment, the first big decision is still a gravity call. API (American Petroleum Institute) separators are long, shallow tanks engineered so buoyancy does the work: free oil rises, heavy solids sink, and a middle, mostly‑water layer passes on. The physics is Stokes’ law (buoyancy‑driven rise/settling for small spheres), with oil droplets (density about 0.8–0.9) floating and solids (density >1.0) settling under quiescent flow (watermanaustralia.com).

In practice, these units are designed to remove droplets ≥0.015 cm (150 μm); smaller droplets remain and need downstream flocculation or flotation (watermanaustralia.com) (scribd.com). As the primary stage in many trains, they sit alongside other primary treatment systems — think screens and oil removal — that are packaged under wastewater physical separation.

Gravity separation limits and oil forms

Typical refinery influent blends three oil types. Free oil separates quickly under quiescent conditions and skims off in the API (watermanaustralia.com). Emulsified oil — fine droplets about 1–20 μm stabilized by surfactants — does not separate by gravity and needs chemical or flotation treatment (thewastewaterblog.com). Dissolved oil (molecules truly in solution) is not removable by separation devices and requires biological or advanced treatment.

Translation for operators: a well‑run API separator removes the majority of free oil and coarse suspended solids, but not emulsified or dissolved fractions. Plants that chemically destabilize emulsions often look to coagulants and flocculants before flotation, staying within the paper’s scope of “chemical or flotation treatment.”

Geometry and hydraulics per API 421

API Publication 421 and industry practice converge on a simple shape: long, narrow, and shallow to approximate plug flow (near‑uniform velocity profile) and minimize rise distance. Key geometry: length‑to‑width ratio ≥5:1; depth‑to‑width ratio about 0.3–0.5 so droplets have a short rise path (watermanaustralia.com). Practical limits land at channel width ≈20 ft and depth ≈8 ft to keep chain mechanics manageable and flow laminar (watermanaustralia.com).

Hydraulically, designers cap horizontal velocity at ≤3.0 ft/min (≈0.015 m/s) to sustain laminar flow and residence time; higher velocities induce turbulence and short‑circuiting (watermanaustralia.com) (thewastewaterblog.com). One designer note is explicit: keeping Vh≤3.0 ft/min reduces turbulence and its influence on blocking oil separation (watermanaustralia.com). Reaction‑inlet baffles or diffusion devices are usually installed at the inflow to distribute wastewater and prevent short‑circuiting (not easily cited, but standard practice).

Mechanically, chain‑and‑flight scraper conveyors on the bottom continuously remove settled sludge, while a rotating surface oil skimmer (“oil‑roll”) collects floating oil (watermanaustralia.com) (scribd.com). Many separators are divided into two parallel channels (API 421 recommends n=2) so one can run while the other is serviced.

Sizing calculations and rise velocity

Sizing convention follows API 421 (Hazen’s equation reference). Start with design flow Q (ft³/min) and divide by the chosen horizontal velocity Vh to get needed cross‑section A = Q/Vh (ft²). Typical practice sets Vh to 15×Vt, where Vt is the rise velocity of a 150‑μm oil droplet, but never exceeds 3 ft/min (sumiowater.com).

At 20 °C with water specific gravity ≈0.991, oil ≈0.858, and viscosity μ≈0.0103 poise, Stokes’ law yields Vt≈0.31 ft/min for a 150‑μm droplet (sumiowater.com). Using Vh=3 ft/min gives A = Q/3. Designers then apply a turbulence factor F≈1.2–1.5 to size length L = F·Vh/Vt·d (d is depth) and verify L/B≥5. In practice, select channel width B (minimum 5–7 ft for scrapers) and solve for d and L to meet A and ratios (sumiowater.com). The published example shows Vt=0.31 ft/min, Vh=3 ft/min, and A_c=Q/Vh=4.9 ft² for a small flow (sumiowater.com). The bottom line is an oversized cross‑section (large B·d and often multiple channels) to achieve the low velocity necessary for 150‑μm droplets to rise.

Performance ranges and polishing steps

In the field, a well‑designed API unit typically removes 80–90% of free oil and a significant fraction of total suspended solids (TSS), yet effluent still carries oil in the tens of mg/L. Industry targets often set ≤200 mg/L oil in API effluent, with about 50 mg/L achievable in many cases (thewastewaterblog.com). Reported ranges widen with conditions: oil removal 52–99%, chemical oxygen demand (COD) removal 5–45%, TSS removal 10–68% (thewastewaterblog.com) (thewastewaterblog.com).

To meet stringent discharge limits, a polishing step — for example, a corrugated‑plate interceptor or dissolved‑air flotation (DAF) — is generally added. Plants commonly deploy DAF systems downstream, especially when upstream chemical destabilization is part of the plan described above.

Operating discipline: flow, pH, temperature

Three variables dominate daily performance. Flow must remain at or below design (velocity ≤3 ft/min); exceeding this causes short‑circuiting and a steep efficiency drop. One expert summary is blunt: staying “close to, or below, design flow (velocity … ≤3 ft/min) is critical for optimal performance” (thewastewaterblog.com). pH matters because low pH helps break emulsions; reducing pH below 7 is known to enhance free‑oil separation, while high pH tends to stabilize emulsions (thewastewaterblog.com) (watertechonline.com). Temperature affects viscosity: warmer water improves droplet rise; cold refinery effluent slows separation.

Where dissolved oil must be handled, operators look beyond separation to biological or advanced treatment. That is the space covered by biological digestion systems, which sit downstream of primary and flotation steps.

Monitoring and regulatory context

Real units rely on weirs and samples to verify performance, and it is common to see API effluent oil in the 10–100 mg/L range. For context, U.S. EPA pretreatment standards for new refineries target less than 10 ppm oil in final effluent (nepis.epa.gov) — well below what an API stage alone achieves, hence the need for polishing.

Indonesia’s rules sit in the same band: a 1996 Ministerial Decree set the oil/grease limit at 20 ppm for refinery wastewater (journal.lemigas.esdm.go.id), and Permen LH 19/2010 specifies 25 mg/L oil+grease for produced and drainage waters from onshore oil/gas operations (id.scribd.com). The figures are comparable to many country standards (10–30 mg/L range), meaning the API must remove most free oil if effluent is to meet discharge limits.

Maintenance and sizing discipline

Sizing for worst‑case conditions is non‑negotiable. Undersizing — too much flow for too little area — drives turbulence and short‑circuiting. If actual flow exceeds design by 50%, velocities jump and droplets can’t rise in time; hence design for peak/surge flows (often 1.2–1.5× average) is essential.

Housekeeping is equally determinative. Oil and solids must be removed continuously via the rotating surface skimmer and chain‑and‑flight bottom scraper (watermanaustralia.com). Letting a surface scum layer thicken beyond about 10–20% of tank depth shrinks effective volume and traps new oil; accumulated sludge does the same on the bottom. In practice, operators often remove sludge daily or weekly depending on load, and covered (enclosed) API tanks may require inerting to prevent vapor release. Routine skimming, sludge dredging, and baffle checks are the difference between meeting targets and breakthrough.

Supporting gear — skimmers, scrapers, level and oil‑in‑water sensors — is standard fare among wastewater ancillaries and helps keep separators on‑spec.

Measured outcomes and what they mean

When designed and maintained as above, an API stage routinely drops oil by an order of magnitude: influent at several thousand mg/L oil & grease may exit at a few hundred mg/L or less. EPA examples show influent oil at 100–300 ppm reduced to 20–40 ppm in a single API stage — that’s 60–90% removal (nepis.epa.gov). Removing more than 90% of a 300 ppm feed yields roughly 30 ppm effluent — still above strict limits and therefore a signal for downstream polishing.

The broader ranges cited earlier — oil 52–99%, COD 5–45%, TSS 10–68% (thewastewaterblog.com) — reflect well‑run systems. Deviations almost always trace back to poor sizing or lax maintenance.

Key takeaways and source notes

The through‑line is consistent: engineer to API 421’s long‑shallow‑low‑velocity template, size for peaks, and run the skimmers and scrapers like clockwork. The data — from Stokes’ law rise calculations (sumiowater.com) to field reports (thewastewaterblog.com) — underline that proper sizing and housekeeping are as vital as initial design for meeting environmental goals and protecting downstream treatment.

Industry references cited throughout include Waterman Engineers’ design notes on geometry, velocities, and operations (watermanaustralia.com) (watermanaustralia.com) (watermanaustralia.com) (watermanaustralia.com) (watermanaustralia.com); the USFilter/WastewaterBlog technical explainer on emulsions and performance ranges (thewastewaterblog.com) (thewastewaterblog.com) (thewastewaterblog.com); sizing math examples and constraints (sumiowater.com) (sumiowater.com) (sumiowater.com); EPA engineering expectations (nepis.epa.gov); and Indonesian regulatory tables (journal.lemigas.esdm.go.id) (id.scribd.com) and maintenance mechanics notes (scribd.com) as well as pH control practices (watertechonline.com).