The refinery bottleneck you can’t see: stopping CDU overhead corrosion before it starts

Overhead corrosion in the atmospheric crude distillation unit is a well-documented refinery problem tied directly to acid formation. As operations lean into heavier and saltier feeds, risk goes up: salts such as NaCl, MgCl₂, and CaCl₂ carried in the crude hydrolyze in the preheat/furnace at roughly 350–450 °F to release HCl (hydrochloric acid) (patents.justia.com). When the HCl-laden vapor cools and water condenses at the acid dew point (the temperature where acidic water begins to condense from vapor), the initial condensate can register “pH’s as low as one or even fractional” (patents.justia.com).

Hydrogen sulfide (H₂S) and organic acids (such as acetic and propionic) compound the damage; HCl catalyzes FeS formation in condensate (patents.justia.com). The attack often concentrates at the initial condensation point (ICP, the very first location where water begins to condense) or at shock points, and also under deposits formed when ammonium/amine salts condense above the dew point (patents.justia.com) (patents.justia.com).

Industry practice puts an overhead condensate chloride target below about 10 ppm to limit corrosion; in one Shell study that frequently used the 10 ppm limit, it was exceeded 60% of the time and the average was ~18 ppm, tracking accelerated carbon-steel corrosion (researchgate.net). In practice, refineries aim to desalt crude to keep overhead chlorides in the low single‑digit ppm range (e.g., <5–10 ppm).

Multifaceted control program design

The backbone is optimized desalting upstream. Crude is washed with fresh water—typically 3–5% by volume—in one or more stages to dissolve and coalesce salt droplets. A two‑stage desalter is considered best practice for heavy/sour crudes. Tosco’s Aurus refinery processing heavy 13°API California crude achieved ~95% salt removal in two stages, leaving less than 1 lb salt per 1,000 bbl of crude (~3 ppm) (ogj.com). A single wash “can remove up to about 90% of the salt” and additional stages are commonly used on heavier crudes (patents.justia.com).

If required, caustic (NaOH) injected just downstream of the desalter converts MgCl₂/CaCl₂ into more thermally stable NaCl, but the treat rate is constrained by downstream fouling and catalyst poisoning limits (patents.justia.com). In practice, refiners tune mixing energy, water-to-oil ratio, demulsifier dosage, two‑stage separation, and even pluggable pipeline settling tanks to keep HCl precursors low. Demulsifier supply and selection sit within oilfield chemical programs often managed alongside other oilfield chemicals.

Real-time overhead monitoring and probes

In the CDU overhead itself, control is chemical plus analytical. Refineries rely on inline ER (electrical resistance) corrosion probes (devices that measure metal loss through resistance change), ultrasonic wall‑thickness monitors, corrosion coupons (removable metal specimens), and routine sampling of overhead condensate for pH and chloride. A Shell program logged unit duty, temperature/dewpoint profiles, desalter efficiency, and overhead drum chemistry (pH, Cl⁻, NH₄⁺, Fe) while tracking on‑line corrosion sensors to connect operating state with corrosion response (researchgate.net) (researchgate.net).

Neutralizing amines and pH targets

Neutralizing amines (water‑soluble bases) and/or ammonia are used to hold overhead condensate pH in the roughly 5.5–6.5 range. Guidance from Nalco/Exxon points to ~5–6.5 (some target 5.5–6.0) (patents.justia.com) (patents.justia.com). Common neutralizers include ethanolamines, morpholine, cyclohexylamine, monoethanolamine (MEA) or more volatile bases like monoethanolamine (MEA) and N,N‑dimethylethanolamine (MDEA), diisopropanolamine, among others. Ammonia (gas or aqueous NH₄OH) remains a workhorse: it neutralizes HCl to NH₄Cl and readily vaporizes; bulk overhead ammonia originates partly from the crude and partly from deliberate addition for pH control (patents.justia.com).

Results over a full year in one industrial unit showed that varying neutralizer rate moved condensate pH only modestly—pH stayed ≈5–8 almost regardless of rate (researchgate.net). Overdosing is risky: excess amine with HCl can form solid ammonium/amine hydrochlorides above the dew point, triggering “dry” under‑deposit corrosion (researchgate.net) (patents.justia.com). Instances of salt deposition above the dew point have increased with heavy amine dosing, particularly as crude slates got heavier (patents.justia.com).

Placement matters. Injection is often 5–10 trays below the column top or into the top condenser feed—before final condensation—so the receiver drum pH holds in range (patents.justia.com) (patents.justia.com). Operators adjust dosing to maintain 5.5–6.5 while watching for NH₄Cl crystals; the neutralizer blend—often ammonia with higher‑boiling amines like MEA or MOPA—is tuned to maximize pH lift with minimal salt volatility (patents.justia.com) (patents.justia.com). Lab‑scale overhead simulators (NEUs) have been used to screen blends that achieve target pH at given chloride loads (patents.justia.com) (patents.justia.com).

Measured outcomes align: keeping condensate near pH ~6 versus ~4 can reduce corrosion rate by about an order of magnitude, since acid attack is driven by [H⁺]. Plants have recorded near‑zero rates on corrosion probes once overhead water pH is stabilized at 5–6.5. One EU comparison found a Nitro brand amine delivered ~70–90% lower metal loss rates than plain ammonia at similar chloride. Modern setups sometimes automate dosing on continuous drum‑pH feedback. In practice, neutralizer supply is often specified as a neutralizing amine, and injection is metered with accurate chemical dosing equipment such as a dosing pump.

Film‑forming amines for barrier protection

“Filming” corrosion inhibitors complement neutralizers by protecting metal surfaces from intermittent wetting. These are water‑insoluble fatty amines or diamines that adsorb onto steel through their amine group (R–NH₂) and create a hydrophobic organic layer roughly 10–20 Å thick; hydrocarbon chain lengths are typically C₈–C₂₂ (researchgate.net). Commercial FFAs include octadecylamine (ODA) and N‑oleyl‑1,3‑propanediamine (OLDA) (researchgate.net).

They’re usually dosed as oil‑soluble concentrates or emulsions into hot overhead lines or reflux so they vaporize and condense with naphtha/water. Nalco/Exxon notes they “protect equipment by forming a barrier on the steel surface” but are not highly effective right at the ICP (chlorides and acidity are extreme and liquid film is minimal) (patents.justia.com). Typical programs maintain 5–20 ppm of FFA relative to condensate water, with some units pulsing higher on upset. Field data point to benefits: several refineries saw ~50–90% drops in overhead copper/steel corrosion rates; one lab result cut iron corrosion current density from ~30 nA/cm² to under 3 nA/cm² with 10 ppm ODA in condensate. Because volatility is low, about half stays in hot wells and the rest evaporates, so re‑dosing is periodic. In many programs the filming package is specified as a corrosion inhibitor for overhead service.

Program integration and KPIs

A complete program links desalting, neutralizer, filming, and monitoring with clear triggers. Typical KPIs: overhead water chloride below ~5–10 ppm and condensate pH 5.5–6.5. Online pH and conductivity, plus periodic lab titration/ion analysis, track overhead chemistry; corrosion coupons or probes at elbows, exchangers, and the receiver provide metal‑loss data.

One refinery’s long‑term data showed that when neutralizer was reduced and pH drifted from 6.5 to 4.5, coupon corrosion jumped from ~0.01 mm/y to ~0.1 mm/y (researchgate.net) (patents.justia.com). Adding a small amount of FFA eliminated erratic spikes in the overhead exchanger. The recommended action matrix is consistent: maintain optimum desalter operation, then adjust overhead amine treatments only enough to hit pH targets and avoid dry‑salt formation; heavier crudes or lower distillation temperatures call for tighter controls.

Desalting as first defense

Because chlorides are the root cause, desalting efficiency is paramount. Best practice is a two‑stage electrostatic desalter with robust mixing and chemical demulsifiers, aiming for ≤10 ppm Cl⁻ in desalted crude (patents.justia.com) (ogj.com). The rule of thumb is simple: each 90% removal step drops residual salt by roughly an order of magnitude—e.g., 1000 ppm feed to ~100 ppm after one wash, ~10 ppm after a second. Tosco’s two‑stage case left less than 1 lb/1000 bbl (about 3 mg/L chloride) (ogj.com).

The downstream impact is material. Every 10 ppm drop in overhead chloride reduces HCl partial pressure substantially. Plants that sharpened two‑stage salt removal—through better mixing and chemicals—reported 80% of previous overhead pipeline leaks were prevented. Modern strategies typically achieve under 10 ppm net crude chlorides on light crude and a few ppm on sweetened crude (ogj.com) (patents.justia.com), cutting potential HCl generation in the steam drum/water seal by ~90%.

Documented targets and sources

Inline references support each element: >90% salt removal in staged desalting (patents.justia.com); maintaining overhead chloride under ~10 ppm (researchgate.net); and condensate pH targets around 5.5–6.5 (patents.justia.com) (patents.justia.com). The cited body includes Schempp et al. (Shell, EuroCorr 2017) and Nalco/Exxon (U.S. Patent No. 5,965,785, 1999), alongside real‑world plant cases and published reports (e.g., patents.justia.com; ogj.com).

In implementation, many sites standardize the chemical toolkit—a neutralizing package for pH control (neutralizing amine), a filming package for surface protection (corrosion inhibitor), and precise injection hardware (dosing pump)—and let the data decide the rate. The program focus is consistent: measure, hold pH ~5.5–6.5, keep overhead Cl under ~10 ppm, and maintain a film where it counts.