Pitting, Crevice and Uniform Corrosion

Why this matters

The ugliest corroded part in a building is usually not the one about to fail. Corrosion spread evenly over a large area removes metal slowly and tells you exactly how fast it is going. Corrosion concentrated at a few points removes very little metal, looks like almost nothing, and goes through the wall. Techs who judge remaining life by appearance replace the wrong parts and leave the dangerous ones in service, and the customer's confidence goes with the first surprise leak in something that looked fine last month.

The whole distinction is localization: how much of the surface is doing the corroding. Everything else follows from that one number.

Three shapes of attack

Uniform (general) corrosion removes metal at roughly the same rate everywhere the environment touches. Plain steel rusting in damp air is the standard case. It is unattractive and it is predictable, which makes it the easiest kind to manage: measure the loss, divide by the elapsed time, and you have a rate you can project.

Pitting concentrates the entire attack into small discrete points while the surrounding surface stays passive and bright. It is the normal failure mode of the alloys that protect themselves with a passive film, so stainless steels and aluminum alloys are the usual victims, and chloride is the usual trigger. Once a pit starts, the chemistry inside it becomes more aggressive than the bulk fluid, so the pit accelerates itself.

Crevice corrosion is the same self-accelerating chemistry, but the trigger is geometry rather than a random flaw. A tight gap holds fluid that cannot exchange with the bulk, oxygen inside the gap is consumed and not replaced, and the resulting difference between the shielded and open surfaces drives attack inside the gap. The classic sites are under gaskets and washers, inside threaded joints, in lap joints, under insulation, and under a deposit or a piece of debris sitting on an otherwise clean surface.

The gap size matters and it is counterintuitive: crevice attack favors tight gaps, on the order of a few thousandths of an inch, because a wide gap flushes and never sets up the concentration difference. Loosening a joint to "let it breathe" is not a fix; eliminating the crevice or sealing it completely is.

Why average thickness lies

An ultrasonic thickness gauge reports the wall it is sitting on. That is exactly the right measurement for uniform corrosion and exactly the wrong one for localized corrosion, because the odds of parking the probe on a pit are poor and the surrounding metal is close to nominal.

The relationship between the two is captured by the pitting factor: the deepest penetration divided by the average penetration over the same period. A pitting factor near 1 means the attack is uniform. A pitting factor of 10 or 30 means the deepest point is losing metal ten or thirty times faster than the average, and the average is worthless as a predictor. This is the single most useful number in the topic, and computing it takes one thickness reading and one pit depth.

One measurement pair, two verdicts

The gate is a single comparison, made per component: measure general wall loss with a thickness gauge, and separately measure the deepest single defect with a pit gauge or a depth micrometer. The unit of analysis is the individual component, not the system, because two pipes in one room can sit on opposite sides of this line. Project remaining life from the deeper of the two penetrations, never from the average, and if the pitting factor is above roughly 5, treat the general rate as irrelevant to the decision.

Case A: the part that looks finished

A carbon steel component, nominal wall 0.250 in, installed 12 years ago, covered in layered orange rust with scale flaking off in the hand. The customer wants it replaced on sight.

Cleaned and gauged, the thinnest of several readings is 0.210 in. General loss is 0.250 minus 0.210, so 0.040 in over 12 years, a rate of about 0.0033 in per year. There are no discrete pits: the surface is rough but the variation between readings is small, so the pitting factor is close to 1 and the average is a fair predictor. Against a minimum required wall of 0.150 in for the service, the remaining allowance is 0.210 minus 0.150, or 0.060 in, which at 0.0033 in per year is about 18 more years.

Verdict: clean it, address the water source, coat it, and schedule a re-measurement rather than a replacement. Two caveats belong with that number rather than after it: the 18-year projection assumes the rate stays linear, which is usually conservative for general rusting because the accumulating scale slows further attack, and it assumes the minimum required wall came from the manufacturer or the applicable code rather than from a rule of thumb.

Case B: the part that looks fine

A stainless component in the same building, also nominal 0.250 in wall, also 12 years old, bright and clean apart from a few rust-colored weeping spots the size of a pencil tip.

Gauged between the spots, the wall reads 0.244 in. General loss is 0.006 in over 12 years, a rate of about 0.0005 in per year, which projects to essentially forever. A pit gauge on the deepest weeping spot reads 0.180 in.

Now the arithmetic changes character. Remaining wall at that pit is 0.250 minus 0.180, or 0.070 in. The pitting factor is 0.180 divided by 0.006, which is 30. The pit's own penetration rate is 0.180 over 12 years, about 0.015 in per year, so the wall left at the pit is 0.070 divided by 0.015, under 5 years, and that is the optimistic version because pits accelerate rather than run linear.

Verdict: this is the part to replace, and the cause has to be found before the replacement goes in, because an identical part in the same chloride-bearing environment starts the same clock. Note that the pit rate and the general rate above are not the same kind of measurement and should not be compared as if they were: 0.015 in per year is penetration at one point, 0.0005 in per year is loss across the surface, and only the first one predicts a hole.

The pair together is the lesson. The part that looked finished has roughly 18 years. The part that looked fine has under 5. Appearance ranked them backwards, and one pit gauge reversed the call.

What each shape responds to

Two of those "do not" entries are worth their own sentence. Cleaning pitted stainless makes it look repaired while leaving every pit as an active site that restarts immediately. And a crevice cleaned but reassembled the same way is a crevice again the moment the joint is torqued.

What changes the answer

• Stagnation. Most localized attack needs still fluid. A system that runs continuously suppresses pitting and crevice attack that appears within months of the same system being left idle over a season. If failures started after a shutdown or a schedule change, that is the finding.
• Temperature. Localized attack on passive alloys accelerates with heat, and above roughly 60 C (140 F) with chlorides and tensile stress present, the failure mode can shift from pitting to stress corrosion cracking, which gives even less warning.
• Deposits. Anything sitting on a surface creates a crevice underneath it. That is why a system with scale or biofilm suffers localized attack that the same system, clean, does not.
• Which alloy is actually installed. The pitting resistance of stainless grades differs substantially, and a substitution made at a counter is invisible after installation. Verify the grade before concluding the environment is at fault.

Verifying a remaining-life call

• Take multiple thickness readings, not one. Grid the area and record the minimum, not the average, because a single reading cannot tell you whether the surface is even.
• Search for the deepest pit rather than measuring the first one you see. Pit depth distributions are skewed; the one that matters is the outlier. ASTM G46 sets out how to examine and rate pitting, including depth measurement, and is the right thing to cite in a written finding.
• Calibrate the gauge on a known thickness. A thickness reading taken through heavy scale or without a couplant reads low or refuses to read at all, and a low reading condemns a sound part.
• Get the elapsed time right. Every rate above divides by service years. If the component was replaced at some point and nobody recorded it, your rate is wrong by whatever the error is, in the dangerous direction if the part is younger than you assumed.
• Re-measure the same spots, marked. A rate from two measurements at the same location beats any single measurement plus an assumption about the original wall, and it is the only way to catch a rate that is accelerating.

References

• ASTM G46, standard guide for examination and evaluation of pitting corrosion, for pit depth measurement and rating including the pitting factor
• Manufacturer documentation or the code in force for the minimum required wall thickness of a specific component in a specific service; a remaining-life projection is only as good as that number
• See related: How to Read Corrosion and Name Its Cause; How Each Common Metal Tells You It Is Failing; The Water Chemistry That Attacks a System