Refineries rely on a steam‑driven distillation column to strip hydrogen sulfide and ammonia from wastewater. The real performance levers are blunt but decisive: steam rate and feed pH — and getting them wrong is expensive.
Refinery sour water — the wastewater rich in dissolved hydrogen sulfide (H₂S) and ammonia (NH₃) from units like cokers, FCC, and hydrotreaters — routinely arrives loaded with several thousand ppm of NH₃ and up to a few thousand ppm of H₂S in solution, with both species strongly dissolved (H₂S forms bisulfide in alkaline solution; NH₃ dissolves as NH₄⁺) (digitalrefining.com) (climate-policy-watcher.org). The workhorse fix is a sour‑water stripper — a distillation column (stripper) that uses steam to drive off those gases — aimed at bottoms water with NH₃ in the tens of ppm (for example, 30–80 mg/L) and H₂S near zero (often <0.1 ppm) (digitalrefining.com).
Regulatory targets are pushing hard too. Indonesian effluent standards, for instance, limit sulfide to ~1 mg/L and ammonia‑nitrogen to 5–10 mg/L depending on discharge rules (id.scribd.com). Refineries increasingly recycle treated sour water as wash water, further raising the quality bar.
Sour water composition and targets
Non‑phenolic sour water typically contains several thousand ppm NH₃ and up to a few thousand ppm H₂S, with strong solubility effects: H₂S equilibrates as bisulfide under alkaline conditions, and NH₃ as ammonium (NH₄⁺) (digitalrefining.com) (climate-policy-watcher.org). A well‑run stripper (vertical distillation column) aims to cut NH₃ to a few tens of ppm (e.g., 30–80 mg/L) and reduce H₂S to near zero (often <0.1 ppm) (digitalrefining.com).
Column design and operation
The typical unit is a steam‑heated distillation column equipped with trays (bubble‑cap, sieve, or valve) or packing. Feed sour water is preheated (often via a bottoms exchanger) and introduced mid‑column to maximize mass transfer. Live steam, often 50–150 psig (pounds per square inch gauge), is injected in the reboiler or lower section to heat bottoms to near boiling (~120–130 °C), generating rising vapor that carries H₂S and NH₃ upward while most water flows down (digitalrefining.com).
Modern strippers commonly use 35–45 actual trays (with anti‑fouling features), and structured packing is applied in relatively clean feeds (digitalrefining.com). Overhead vapors, rich in NH₃ and H₂S, are condensed or routed to an acid‑gas plant; stripped bottoms are cooled for reuse or discharge. Design practice often favors simplicity: operators have eliminated feed preheaters and reboiler circulation in favor of direct steam injection when trays are reliable, and one field case replaced a conventional reboiler with open steam injection without losing performance (ro.scribd.com) (ro.scribd.com).
Typical steam rates land at 1.0–1.5 lb of steam (50 psig equivalent) per gallon of sour water — about 260–390 kg/m³ — and a benchmark 40‑tray design using ~150 psig steam can heat bottoms to ~128 °C (262 °F) and reduce feed ammonia to ~50 ppm (digitalrefining.com) (brewiki.org). Strippers often run steam at higher pressure than amine regenerators, helping minimize exchanger area and avoid ammonia degradation issues (digitalrefining.com).
Stripping mechanism and energy duty
Steam stripping works by heating to boil, shifting acid–base equilibria (NH₄⁺ → NH₃↑ and HS⁻ → H₂S↑), and diluting acid‑gas partial pressures in the vapor; injected steam carries off NH₃ and H₂S, pushing equilibrium toward the gas phase (digitalrefining.com). Volatility is not equal: at 100 °F (38 °C) and 1 atm, Henry’s law constant (a measure of gas solubility) for NH₃ is ~38,000 ppm/psia vs. ~184 ppm/psia for H₂S — making H₂S far easier to strip (climate-policy-watcher.org). In practice, >90% H₂S removal occurs even at ~100 °F, whereas >90% NH₃ removal typically requires bottoms near 110–120 °C; thus H₂S removal is easier (≫90% achievable) than NH₃ (often ~50–90% unless energy is pushed) (climate-policy-watcher.org) (digitalrefining.com).
Energy use clusters around 1–1.5 lb steam/gal — about 0.25–0.35 MMBtu per ton of sour water — with one industrial design reporting 0.345 MMBtu steam (≈363 MJ) per US ton of feed (digitalrefining.com) (brewiki.org). Many columns rely on thermosyphon reboilers (natural circulation via density differences) or direct steam injection, with practical reboiler temperature limits around ~400–450 °F (~200–230 °C) to avoid corrosion and solids deposition (digitalrefining.com). With overall tray efficiencies typically 25–45%, deep NH₃ removal usually needs 35–45 actual stages, though side‑draw reflux or two‑stage columns can push efficiency higher (digitalrefining.com).
Operating levers: steam rate and feed pH
Turning up steam improves stripping but with diminishing returns. In one refinery simulation, boosting steam cut bottoms H₂S from ~0.48 ppm to 0.15 ppm and NH₃ from ~50 ppm to ~29 ppm, against nominal requirements of <10 ppm H₂S and <100 ppm NH₃ (pmc.ncbi.nlm.nih.gov). The same exergetic analysis showed that increasing steam from ~74,000 to 76,200 lb/hr improved purity but dropped global exergy efficiency (a measure of useful energy) from ~44% to ~36% (pmc.ncbi.nlm.nih.gov). In practice, operators increment steam to hit bottoms NH₃ around 30–50 ppm and H₂S <0.1 ppm (digitalrefining.com) (pmc.ncbi.nlm.nih.gov).
pH control is equally pivotal. Acidity/alkalinity shifts speciation: making feed acidic “fixes” ammonia as NH₄⁺ and frees H₂S to strip; adding caustic converts NH₄⁺ to volatile NH₃ but also converts H₂S to non‑volatile HS⁻. Two‑stage designs exploit this, often acidifying the first stage to remove H₂S and adding caustic in the second to remove NH₃ (climate-policy-watcher.org). In single columns, modest caustic addition can help NH₃ removal; note that inefficient stripping often shows up as high‑bottoms pH because H₂S leaves first, leaving NH₄⁺ behind (ro.scribd.com). Plants typically meter acids/caustics with precise dosing pumps to hold target pH.
Because upstream sources can be strongly acidic or basic, operators often track source pH and partially neutralize before stripping. Some maintain bottoms pH around 8–10 to maximize NH₃ drive‑off, accepting that H₂S removal may then need more steam or an additional stage. Overall efficiency is maximized by balancing steam duty and pH: sufficient steam (or lower operating pressure) plus tuned pH keeps H₂S and NH₃ below limits (climate-policy-watcher.org).
Performance benchmarks and design trends
Optimized units routinely deliver >95–99% H₂S removal and 80–90% NH₃ removal. Historical data show >90% H₂S removal at ~100 °F, while ~90% NH₃ removal typically needs bottoms above ~110 °C (climate-policy-watcher.org). Bottoms ammonia slips of 30–80 ppm are common (digitalrefining.com). To improve removal, sites add trays, increase steam, raise column height, or run a second column (climate-policy-watcher.org) (digitalrefining.com). The trendline points toward higher pressures and robust designs: 35–45 stages with direct steam injection and durable trays are now typical to reach <30 ppm NH₃ effluent (digitalrefining.com).
Regulatory context and control strategies
Meeting discharge demands often sets the operating point. Indonesian limits of ~1.0 mg/L sulfide and 5–10 mg/L ammonia (as N) require high‑end efficiencies (id.scribd.com). Many refineries recycle stripped water internally, or send it to biological treatment, where maintaining some ammonia can be beneficial; conventional activated sludge systems are a common downstream step.
Operators monitor steaming duty and adjust during upsets. Closed‑loop control of reflux or steam/feed, together with pH and temperature sensors, is standard practice; one application note highlights stabilizing overhead composition via automated steam/feed control (yokogawa.com). Overhead pressure is typically held at 1–30 psig to optimize condensation. Feed pH is sampled and sometimes adjusted with caustic only if ammonia removal lags; acid addition (e.g., CO₂) is used only if H₂S removal is the bottleneck. Plants backstop these control loops with supporting equipment sized for duty.
The bottom line on energy and specs
In single columns, typical performance is ~1.0–1.5 lb steam/gal (50 psig), residual NH₃ ≈30–50 ppm, and H₂S ≪1 ppm (digitalrefining.com) (digitalrefining.com). An exergy study pegged base‑case overall exergy efficiency near ~44%, dropping to ~36% when steam is raised to chase lower NH₃/H₂S (pmc.ncbi.nlm.nih.gov). It is the central trade‑off: more steam and more stages can meet stricter limits but at higher energy cost. With adequate steam and ~40 trays, up to ~99% H₂S and ~90% NH₃ removal are achievable (climate-policy-watcher.org) (digitalrefining.com).
Summing up the operating prescription: a well‑designed stripper uses sufficient steam — typically 0.25–0.35 GJ per m³ of water — and enough stages (35–45 trays), then fine‑tunes steam and pH until gains level off (digitalrefining.com) (digitalrefining.com). Data‑driven control delivers compliance — for example, <1 mg/L H₂S and <10 mg/L NH₃ in discharge (id.scribd.com) — while conserving energy and reagents.
