CO₂ (Sweet) Corrosion: Chemistry, Drivers, and Field Signatures

The gas analysis comes back from a wet-gas gathering line you sampled last week. You scan down the composition table — methane, ethane, propane, the usual — and there it is, three rows from the bottom: carbon dioxide, 2.1%. The operator reads it over your shoulder. “Couple percent CO₂ — I’ve got another line carrying more than that and it’s never given us a bit of trouble. This one’s probably fine too, right?”

Fair question — and here’s the honest answer this module is built around: you can’t tell from that number alone, and neither can he. That 2.1% might be a line that runs clean for thirty years, or a line that’s quietly losing wall right now. The operator isn’t wrong that he’s seen higher CO₂ run clean — he’s just missing the second number that explains why. It’s sitting right there on the line’s paperwork, and once you put the two together, the answer takes about ten seconds of arithmetic.

The module before this one handed you the chemistry frame: four ingredients, one reaction, the levers that make water aggressive, and CO₂ named as the first of the four aggressive solutes. This module takes that first solute end to end — what it does in the water, what number actually measures the threat, what the damage looks like on steel, where it lives in real systems, and what you collect when you suspect it. By the end, you’ll read that gas analysis the way the operator can’t.

What “sweet” actually means

CO₂ corrosion is acid attack — full stop

First, the name. Gas that carries hydrogen sulfide is called sour — the old-timers named it off the smell. Gas without H₂S is sweet. So “sweet corrosion” just means the damage CO₂ does when it’s the acid gas running the show.

And what it does is simpler than its reputation. CO₂ in the gas phase touches nothing — dry CO₂ and bare steel can ride together indefinitely. The trouble starts when there’s liquid water on the wall, because CO₂ dissolves into water and a slice of it becomes carbonic acid. You met this reaction in the last module; here’s the full three-step chain it belongs to:

You don’t need to derive or balance any of that. You need the picture: a CO₂ line with wet walls is an acid factory. Not a strong acid — carbonic acid is mild compared to anything in a lab cabinet — but the line doesn’t get one dose. It gets a fresh supply manufactured around the clock, for years, replenished every time flow brings more CO₂ past the water. Mild acid on an endless drip beats strong acid in a single splash.

One contrast line before we move on, because the next module owns the rest of it: sweet corrosion is acid attack. Sour corrosion — H₂S — is acid attack plus a hydrogen problem inside the steel itself. Plenty of real systems carry both gases at once, and when they do, the H₂S side usually sets the safety stakes. That story is next door. This module is the pure-sweet story, and it’s the right one to learn first because the CO₂ chain underneath is the backbone either way.

The working number: partial pressure

So how much acid does the line make? Here’s where a bare percentage stops being the whole story. The amount of CO₂ that dissolves into the water — and therefore the amount of acid manufactured — doesn’t follow the percentage. It follows the partial pressure.

Why this works is intuitive enough: pressure is what pushes gas into water. Double the system pressure and you double the push behind every molecule of CO₂, even if the percentage never moves. Which means the same gas composition can be harmless in one line and aggressive in another:

• 2% CO₂ in a line at 100 psia → P_CO₂ = 2 psia. Mild.
• The same 2% CO₂ in a line at 1,500 psia → P_CO₂ = 30 psia. Fifteen times the acid-making push — same composition, same steel, dramatically different threat.

That’s why a composition number without a pressure number is half a sentence. Run the arithmetic once, properly:

Get this calculation into your hands until it’s automatic. It’s ten seconds with a phone calculator, it’s the first thing the chemistry cares about, and it’s the difference between reading a gas analysis and just filing one.

The three-zone rule of thumb

With a partial pressure in hand, you can place the line on the scale the industry has used for decades as a first-pass screen for carbon steel:

• Below ~7 psia — sweet corrosion is usually negligible. Not impossible, but rarely the threat that drives the program.
• 7 to 30 psia — the moderate zone. Sweet corrosion is credible and wants active management: monitoring locations, sampling cadence, likely an inhibitor program, attention from the integrity team.
• Above ~30 psia — the high-risk zone. Expect aggressive damage unless something strong is standing in the way — serious mitigation, or steel upgraded to corrosion-resistant alloys.

Two honesty notes. First, these are rules of thumb, not hard lines — a line at 6.5 psia isn’t “safe” and a line at 8 isn’t “condemned.” The zones are a triage tool: they tell you how much attention a line deserves before any deeper work is done. Second, everything else this module teaches — temperature, water chemistry, flow — moves a line’s real behavior up or down within its zone. The partial pressure sets the table; the rest of the conditions decide what gets eaten.

The scale that fights back — and why it’s a fragile friend

Now for the wrinkle that makes sweet corrosion interesting instead of just unfortunate. Look back at the chemistry chain: the attack dissolves iron into the water as Fe²⁺, and step 2 left bicarbonate and carbonate drifting around. Put enough of both in the same water and they find each other:

Here’s the part worth sitting with: the corrosion makes the scale, and the scale can throttle the corrosion. When iron carbonate lays down as a tight, well-adhered layer, it physically separates water from steel, and the attack underneath can drop to a small fraction of its bare-steel rate. Lines have run for years in the moderate zone wearing a stable siderite jacket and barely paying for it.

The catch is in the word stable. That scale only forms — and only survives — inside a window of conditions: warm enough for the mineral to precipitate densely, water chemistry holding enough carbonate nearby, pH not so low that the acid redissolves the scale as fast as it forms, and flow gentle enough not to tear it off the wall. Push any of those out of the window — a line cools through winter, a slug of unbuffered water comes through, velocity steps up after a tie-in — and the jacket stops forming, or starts coming off in patches.

That’s why the wise framing is this: the scale is not the protection — the conditions that keep the scale intact are the protection. A line that “looked great for years” didn’t earn a pass; it was living inside the window. Operating changes can move the window off the line without anyone touching the gas composition at all — and the line that was coasting starts losing wall fast. When a long-quiet line suddenly turns up active, the first question isn’t “what changed in the gas?” It’s “what changed in the conditions?”

The temperature paradox

The last module told you temperature is an amplifier, not a hazard gauge — and that some scales actually get more protective when hot. Sweet corrosion is the headline example, and the result is counter-intuitive enough that it deserves its own section, because learners miss it and the field punishes the miss.

Common sense says hotter = faster chemistry = worse corrosion. For sweet corrosion, that’s only true at the cool end. The real curve rises, peaks, and then falls:

• Cool lines — below roughly 100 °F (40 °C): the acid chemistry runs slowly. Attack happens, but at modest rates. Most buried gathering systems in our part of the world live here much of the year.
• The danger band — roughly 140–175 °F (60–80 °C): the chemistry is running hot and fast, but the wall is not yet hot enough for iron carbonate to lay down as a dense, protective layer. Fast attack, flimsy scale — this is where sweet corrosion rates typically peak.
• Hot lines — above roughly 175–195 °F (80–90 °C): the underlying chemistry is faster still, but now the siderite precipitates dense and tight, and the scale jacket wins. Measured corrosion rates often drop on the hot side of the peak.

One field translation: temperature here means the wall, not the wellhead gauge. A stream that leaves the pad at 130 °F can be down at soil temperature within a few hundred yards of burial — which moves the same line through different parts of this curve along its own length, in different seasons. Where the line sits on the curve is a location-by-location question, and it’s part of why monitoring spots get chosen the way they do.

Same gas, different water

The last module spent a full section on alkalinity — the water’s stock of dissolved bicarbonate, its shock absorber against acid. Sweet corrosion is where that lesson cashes out, because CO₂ is the acid the buffer was born to absorb: the bicarbonate in a mineral-rich produced water soaks up carbonic acid’s H⁺ before the pH can fall far.

The practical consequence is dramatic. At the same CO₂ partial pressure, fresh unbuffered water — condensed water on a cool wall, frac flowback, leftover hydrotest water — can corrode carbon steel five to ten times faster than a well-buffered produced water. Same gas, same pressure, same steel; the water’s armor is the whole difference. When two lines off the same pad behave like strangers, the alkalinity line on the lab report is the first place to look.

And remember which water has no armor at all: the distilled-clean film that condenses out of warm gas onto a cool pipe wall. In a wet-gas line, that film forms along the top of the pipe — which sets up a particular damage pattern called top-of-line corrosion, nasty for reasons that include being out of reach of the usual protections. It gets a proper treatment in its own module later in this set; for now, just connect the dots you already hold: condensed water + CO₂ = full-strength acid, exactly where nobody’s sampling.

What it looks like — the mesa fingerprint

Sooner or later you’ll hold the evidence in your hand — a retrieved coupon, a cut-out spool, an end of pipe at a repair — and the damage pattern itself will tell you which mechanism did it. Every mechanism in this set leaves a different signature on steel. Sweet corrosion’s is distinctive enough to carry its own name.

Mesa attack looks exactly like the name: flat-topped islands — little plateaus of surviving surface where the iron carbonate scale held — separated by sharp-walled, flat-bottomed pits where the scale broke down and the acid worked the bare steel below. Steep sides, abrupt edges, like a desert mesa landscape shrunk onto a pipe wall. It’s the visible record of the fragile-friend story: the scale protected here, failed there, and the boundary between the two is a cliff.

File the contrast cards now, surveyed in the last module and taught deep in their own modules ahead: black, smeared sulfide scale says sour chemistry. Orange-brown tubercles say oxygen got in. Flat-topped islands and sharp-walled pits under a grayish carbonate scale say sweet. None of these is courtroom proof on its own — but each one tells you which chapter of the chemistry to open first.

Where you’ll meet it

Sweet corrosion lives anywhere three things overlap: measurable CO₂, liquid water, and carbon steel. In practice, for techs working this region, that means it’s nearly everywhere you work. Appalachian Basin wet-gas gathering — Marcellus and Utica production — typically runs 0.5 to 3% CO₂, with water dropping out at every cool point in the system and long stretches of buried carbon steel to receive it. Run the partial-pressure math on a few hundred to a thousand psig of gathering pressure and most of those lines land in the active-management zone. Some basins run far richer — gas in parts of the Permian and Anadarko carries 5 to 15% CO₂ — and systems that inject CO₂ on purpose for enhanced recovery handle the threat at industrial strength. But the everyday version, the one on your work orders, is the gathering system making a little acid every hour of every day.

The hot-spot problem

One more field pattern belongs in your sweet-corrosion picture, because it’s the one that embarrasses good programs. A line is inhibited, the bulk samples look healthy, inhibitor residuals measure fine — and then an inspection run lights up a cluster of deep wall-loss patches at an elbow. Everywhere else: clean. “But our inhibitor program looks fine.”

It probably is fine — on the straightaways. What happened at the elbow is local: flow that’s gentle in the long runs gets fast and chaotic where geometry turns or chokes — elbows, tees, weld beads standing proud of the wall. In those spots, the moving fluid can strip away the very things doing the protecting: the inhibitor film, the iron carbonate scale, or both. Strip the armor off one patch of steel while the acid chemistry keeps running, and that patch corrodes at bare-steel rates while the rest of the line coasts. The damage concentrates exactly where the protection is hardest to keep in place.

The pattern goes by flow-induced localized corrosion, and the mechanics of it — how fast is too fast, what shear does to films, where erosion takes over — belong to the erosion-corrosion module later in this set. What this module hands you is the recognition: localized damage clustered at fittings and direction changes, in a line whose bulk numbers look healthy, is a flow story — not a chemistry-program failure. Knowing that sentence keeps a tech from chasing the wrong fix.

From a suspicion to a plan

Put the whole module into the working chain you already know — indirect data, the repeatability gate, direct data, mitigation designed upstream. Sweet corrosion slots into it cleanly; this is what the chain looks like when CO₂ is the suspect.

Indirect data. The gas analysis gives you composition; the operating pressure turns it into partial pressure; the three zones tell you how seriously to take it. The water tells the rest: pH trending down from its baseline, dissolved iron trending up — the receipt that metal is leaving — and the alkalinity line saying whether this water has armor. Solids from pig runs that show grayish carbonate scale fit the sweet story. And the things that aren’t there matter too: no H₂S in the gas, no black sulfide scale, no rotten-egg smell — the absences are what make the case sweet rather than sour.

The repeatability gate. Unchanged from the last module, and worth repeating because it protects everyone’s money and credibility: one aggressive sample is a note in the file; two consecutive reads telling the same story is a signal. Sampling has too many ways to lie for one result to move a program.

Direct data. When the signal clears the gate, coupons go in — placed where this module says sweet corrosion concentrates: where water lays, where the temperature band is worst, downstream of fittings. A retrieved coupon answers what indirect data can’t — how fast metal is leaving (corrosion rate) and whether it’s pitting (pitting rate) — and with sweet corrosion it often comes back wearing the mesa fingerprint under its carbonate scale. Placement and retrieval discipline get their own modules later in the track.

Prediction — and whose tool it is. The industry has prediction models for sweet corrosion — the best-known is called de Waard–Milliams — that take in the very numbers this module taught you to collect (CO₂ partial pressure, temperature, pH) and put out an estimated corrosion rate. Running and correcting those models is the integrity engineer’s work, and the depth of it lives far up the track. Your stake is simpler, and worth taking pride in: the model is only as good as the field data that feeds it — and the tech who delivers clean, repeatable, well-documented numbers is the one whose lines get modeled right.

Mitigation, designed upstream, executed by you. When direct data confirms a real problem, the response comes in three familiar families: chemical — a continuous corrosion-inhibitor program, the workhorse for sweet systems; operational — drying the gas harder, managing velocity and water drop-out; and material — upgrading hot spots or whole segments to corrosion-resistant alloys where chemistry can’t keep up. Which lever, what product, what dose — that’s designed by the integrity engineer or corrosion specialist, taught in the mitigation modules much later in the track. The tech executes the program in the field and keeps the samples coming, because follow-up data is what proves the fix is working. The chain loops; it never really ends.

Back to the gas report

So — the analysis on the hood of the truck. CO₂: 2.1%. The operator figures it’ll run clean like his other line. The line runs at 1,000 psig.

Ten seconds: 1,000 + 14.7 ≈ 1,015 psia. P_CO₂ = 0.021 × 1,015 ≈ 21 psia. Middle of the active-management zone — triple the floor of it. “Two percent” was never the number; twenty-one was.

And now the rest of the module fills in the read. Is there water on the wall? It’s a wet-gas line — yes, and condensing fresh wherever the gas meets cool steel. What’s the water’s armor? Pull the alkalinity line from the last water analysis before guessing. Where does the line sit on the temperature curve — warm danger band, or cool and slow? Buried line, so think soil temperature by season, location by location. Any black scale or sulfide smell saying this is more than a sweet story? No — the analysis shows no H₂S. So: a credible, middle-of-the-zone sweet-corrosion candidate, with the next move straight off the chain — water samples with the full report read, gas re-confirmed, solids checked for carbonate scale, and the same collection again next month, because one read is a note and two are a signal.

That’s the difference one module makes. The operator saw a number he’d seen run clean before and figured this line would match. You saw an acid factory at twenty-one pounds of push — and you know his other line probably carried its CO₂ at lower pressure, which is exactly why it ran clean. Same gas, different push, different answer. You walked away with a collection plan instead of a guess. That’s reading a gas analysis.