How to Tell Whether a Coating Has Failed

Why this matters

A coating that looks bad and a coating that has failed are different findings, and shops routinely charge for the wrong one. Chalky, faded paint on a sound, well-bonded film is a coating doing its job while its surface weathers. A perfect-looking film that has quietly lost adhesion over a whole face will pull off in sheets the moment anything is applied over it, and the overcoat gets blamed. The question is never how it looks. It is whether the film is still bonded, whether the substrate is still sound, and whether something in the environment is going to do the same thing again.

Before you scrape, sand, or tape anything

Assessment is mostly looking, but three of the steps below disturb the film.

• Presume lead on a pre-1978 building. For compensated renovation, repair or painting that disturbs paint in target housing or a child-occupied facility built before 1978, above the rule's minor-repair thresholds, the EPA Renovation, Repair and Painting rule at 40 CFR 745 Subpart E requires a certified firm and contained work practices. Test or presume, and do not dry-scrape or power-sand either way. A tape adhesion test on such a surface is a disturbance.
• Treat an unknown industrial primer as chromate-bearing. Grinding or blasting one can generate hexavalent chromium dust, and hexavalent chromium is an INHALATION carcinogen, so the controlling protection is respiratory and not gloves. Use wet methods or local exhaust extraction rather than a wire wheel in still air, with respiratory protection selected under 29 CFR 1910.1026, plus eye and face protection. Where abrasive prep runs on masonry or concrete, respirable crystalline silica applies as well (29 CFR 1926.1153 in construction, 1910.1053 in general industry), and where the coating may contain lead the worker-protection rule is 29 CFR 1926.62 or 1910.1025, which is a separate duty from the EPA renovation rule that protects occupants, and the glove class named on the safety data sheet for any solvent or stripper, which must be available under 29 CFR 1910.1200.
• On live equipment, isolate before you probe. A blister on an operating pressure vessel is not something to pick at while it is under pressure. Isolate, lock, tag and prove zero at a gauge under 29 CFR 1910.147. If the asset is an electrical enclosure, de-energize under 29 CFR 1910.333(b)(2) and prove dead with the live-dead-live sequence in NFPA 70E-2021, 120.5 before a tool touches it.

Step 1: Name what the coating was for

Failure is defined against a purpose, so establish the purpose before you form an opinion.

• Cosmetic or identification. Chalking and fading are failure. Adhesion is not the point.
• Barrier over steel. Continuity is everything. A small break matters more than a large dull area.
• Sacrificial, such as galvanizing or a zinc-rich primer. Small breaks are covered by the surrounding zinc, so the question is whether the zinc is consumed, not whether it is scratched.
• Lining in contact with a fluid. A single holiday puts the fluid on bare wall at one point, so anything less than continuous is a failure regardless of area.

Skip this and you will condemn sound galvanizing for having scratches and pass a lined tank that has one pinhole.

Step 2: Separate the three kinds of signature

Every visible defect belongs to one of three groups, and the group determines the fix, not the appearance.

Degradation of the film itself, adhesion intact. Chalking, a loose powder on the surface from binder breakdown under ultraviolet exposure. Fading. Checking and crazing, a fine surface network of shallow cracks that does not reach the substrate. Wash, assess adhesion, and overcoat.

Loss of adhesion. Blistering, where the film lifts into domes with the substrate still sound underneath. Flaking and peeling, where it has already let go. Delamination between two coats rather than at the substrate, which you find by looking at the back of a flake: substrate-coloured means the whole system let go, primer-coloured means only the topcoat did. These need removal back to sound material.

Substrate failure showing through. Rust bleed out of a break. Undercutting, where corrosion has crept sideways under an otherwise intact film. Filiform, the thread-like tracks that wander under a film in humid conditions. Pitting visible once the film is off. Here the coating is a symptom and the substrate is the finding.

Step 3: Rate extent instead of describing it

"Some rust" is not a record and will not survive an argument. Standard rating scales exist so two people looking at the same asset a year apart produce comparable numbers.

• Rusting is rated on the ASTM D610 scale, running from 10 down to 0, where 10 is essentially rust-free and each step down the upper part of the scale represents roughly a threefold increase in the area showing rust. Record the grade and the distribution word, because spot rusting in one band and general rusting over a whole face are different problems at the same grade.
• Blistering is rated under ASTM D714 by size number together with a frequency term, so a report reads as a size plus a density rather than as "some blisters."
• Photograph each zone from the same position every time.

Step 4: Test adhesion, and test it in the right places

The tape adhesion test under ASTM D3359 is the field method: cut through the film to the substrate, apply and remove a specified tape, and classify what comes away on a scale where 5B is no removal and 0B is the greatest removal, with the standard giving the intermediate bands. The standard provides a lattice of cuts for thin films and a simple X-cut for thicker ones, and it sets the thickness where you switch.

Three rules about where to put the cuts, and they matter more than the technique:

• Test in a sound-looking area next to a defect, not in the defect. The defect has already failed. What you need to know is whether the film beyond it is still bonded, because that is what decides spot repair against full removal.
• Test each exposure zone separately. One face in full sun, one shaded face, and the bottom band that sits in splash and standing water are three different environments on one asset, and they will not give the same result.
• The test is destructive, so cut where you are going to repair anyway. Never on a lining that is in service, and never on a pressure boundary without the owner's agreement in writing.

Step 5: Look for the condition before you blame the product

A coating that has failed in a band, a stripe, or a ring is reporting an environment. Look up from the failure and find it:

• A vertical stripe of blistering below a fitting is a drip.
• A horizontal band at one height inside a vessel or pan is a waterline.
• Failure at the bottom few inches of everything on a site is splash and standing water.
• Failure at every fastener and nowhere else is a crevice, a coating that never got into the joint, or a dissimilar-metal couple at the fastener.
• Failure only on the sun-facing elevation is ultraviolet, and a coating chosen without a topcoat rated for exposure will do it again.

Recoating without fixing the condition buys the same failure on the same schedule, which is the most common way a coating job gets done twice.

Step 6: Decide per zone, against a stated gate

The unit of analysis is one exposure zone, meaning one face, one elevation, or one immersion band, not the whole asset. Then:

• Spot repair when the adhesion test in sound-looking film adjacent to the defect comes back at the high end of the scale, meaning little or no removal, AND the defective area is localized rather than distributed across the zone. The Boolean is AND: good adjacent adhesion with defects scattered right across the zone still means the zone is going.
• Full removal and recoat within that zone when the adjacent adhesion test shows meaningful removal, because the film beyond the visible defect has already let go and anything applied over it is bonded to something that is not bonded to anything.
• Overcoat when adhesion is sound across the zone and the only finding is surface degradation such as chalking, after washing off the chalk so the new film bonds to the coating rather than to the powder.
• Monitor with a date when nothing has failed. Put the review on the schedule with the photograph and the grade attached, because a grade with no earlier grade to compare it to is a description, not a trend.

Worked: a six-year-old outdoor cabinet

A painted steel equipment cabinet, outdoors, six years in service. Four zones assessed: the top surface, the sun-facing vertical face, the shaded vertical face, and the bottom band, taken as the lowest few inches all round.

What each zone showed.

• Top surface: heavy chalking, no rust, no blisters.
• Sun-facing face: chalking and fine checking, no rust, plus one vertical stripe of blistering running down from a condensate fitting.
• Shaded face: sound, mild fading only.
• Bottom band: rust bleed at several spots, blistering between them.

Adhesion tests, one per zone, cut in sound-looking film. Top surface came back at the high end of the scale with only trace removal along the cut edges. Shaded face came back with no removal at all. Sun-facing face, tested away from the stripe, came back at the high end with trace removal. Bottom band, tested a couple of inches above the nearest rust spot in film that looked perfectly good, came back with heavy removal.

Reading it. Three different findings on one cabinet.

The chalking on the top and sun-facing surfaces is film degradation with adhesion intact, which is the coating weathering rather than failing. Wash and overcoat.

The bottom band is the important result, and it is the one you only get by testing outside the defect. The film two inches above the rust looked sound and was not bonded, so the disbonded area is substantially larger than the visible damage. Spot repair here fails the gate on its first condition, not on area. Anything feathered into that band would be applied over film that is already off the steel.

The stripe on the sun-facing face is a condition, not a coating problem. The adhesion test on that face away from the stripe was fine, so the coating is sound where nothing is dripping on it. Recoating the stripe without fixing the condensate fitting reproduces the stripe.

What the zone approach saved. Treating the cabinet as one asset gives one answer, and the honest one-asset answer is full removal and recoat, because one zone genuinely needs it. Assessed by zone, full removal applies to the bottom band only, the two chalked zones get a wash and an overcoat, and the shaded face gets a date. Washing and overcoating a zone takes a fraction of the hours that stripping it takes, and that difference across three of the four zones is the entire saving.

The failure mode if the adhesion test had been skipped. The rust spots in the bottom band look like a spot-repair job, and they would have been feathered and touched up in an hour. The touch-up would have looked correct on the day, the surrounding disbonded film would have continued to lift, and the same band would come back inside a season with the new patches sitting on flakes. All of that is invisible on a walk-around and obvious from one tape test cut in the right place.

Verifying the call after the repair

Two checks, both scheduled rather than remembered.

At full cure, test adhesion at the boundary. Put one cut across the line where new coating meets old, in a zone you overcoated. If the new film comes away from the old at that boundary, the old surface was not clean enough, and you have learned it on one cut instead of over a whole face next season.

At one year, re-rate the same zones from the same camera positions, using the same scales. What you are looking for is not perfection, it is whether any zone has moved down a grade, and specifically whether the zone you fixed a condition on has stayed clean. A stripe that returns after the fitting was repaired means the fitting was not the source.

References

• ASTM D610, evaluating degree of rusting on painted steel surfaces
• ASTM D714, evaluating degree of blistering of paints
• ASTM D3359, measuring adhesion by tape test
• 40 CFR 745 Subpart E, the EPA Renovation, Repair and Painting rule, for disturbing paint in pre-1978 target housing and child-occupied facilities
• 29 CFR 1910.147 for isolating a pressurized or mechanical path, and 29 CFR 1910.333(b)(2) with NFPA 70E-2021, 120.5 for de-energizing and proving dead
• See related: What a Coating Is Protecting Against; Reading Rust and Corrosion Patterns