Why epoxy repairs fail - and the eight steps that stop it

When a coating or a composite repair lifts within a season, the product is almost never the reason. Three preparation failures account for most of it, and all three are measurable, cheap to check, and routinely skipped because they do not look like work.

The three that cause most failures

Soluble salt

Steel that has been in a marine environment carries chloride. Blasting removes rust and scale; it does not remove salt, and salt is not visible. A surface can pass a visual cleanliness standard perfectly and still hold enough chloride to blister a coating from underneath within months, because the salt draws moisture through the film.

This is the single most common invisible cause, and there is a ten-minute test for it.

Surface profile

Adhesion is chemistry plus a mechanical key. If the profile is too shallow, or the abrasive is worn and producing a rounded rather than angular profile, you are relying on chemistry alone and it will underperform. "It was blasted" is not the same as "it was blasted to profile".

Dew point

Applying onto a surface within a few degrees of the dew point means applying onto condensation you cannot see. In an unventilated tank in winter this is far easier to do than most people expect, and conditions inside a tank move during a shift.

The eight steps

This is the published standard our application procedures refer back to. It is worth attaching to a work pack so the requirement is contractual rather than verbal.

The standards, and why they are named

Specifying "clean to bare metal" is not a specification, because two contractors will interpret it differently and both will believe they complied. Naming the standard makes the requirement checkable:

• SSPC-SP 10 / NACE No. 2 - near-white metal blast cleaning.
• ISO 8501-1, Grade Sa 2½ - the visual comparator a surveyor will actually use.
• SSPC-SP 11 - power tool cleaning to bare metal, the recognised alternative for smaller areas or where blasting is impractical.
• ISO 8503 - how profile is assessed.
• ISO 8502-6 - the Bresle method for soluble salt.

When you cannot blast at all

Inside a ballast tank, alongside cargo operations, or on a live system, blasting is frequently not available. That does not make preparation optional - it changes the method.

SSPC-SP 11 is the sanctioned route: mechanical tools and rotation blaster discs to bare metal, followed by a residue-free degrease. Both of the in-service repairs we have documented were prepared exactly this way, in tanks where blasting and pipe replacement were equally impractical.

What does not get a waiver: the chloride limit and the dew point margin. Those two apply whatever removed the rust, and they are the two worth insisting on even when the job is rushed - precisely because it is rushed.

Or change the product instead

Sometimes the honest conclusion is that the surface cannot be brought to a dry, coatable state within the time available. That is a legitimate finding, not a failure, and the answer is a different product rather than a compromised application: wet-applied and underwater grades exist precisely for surfaces that will not be dry.

Choosing the wrong product for a wet surface and hoping is the other half of the failure statistics.

The cheapest insurance on the job

A Bresle sampler and a dew point meter cost very little against a tank recoat, and between them they catch the two failure modes you cannot see. If a preparation inspection budget only stretches to two instruments, those are the two.