Epoxy floor failure: peeling, blisters, chalk, yellowing

Epoxy floor failure shows up in 5 ways. The one you have decides if you need a recoat or a full tear-off. Sheet peeling, blisters, hot-tire pickup, yellowing (ambering), and chalking each come from a different cause. You can sort yours out in about 5 minutes, standing on the floor with a knife and a flashlight, before anyone quotes you a price.

That matters because three crews can look at the same failed floor and tell you three different stories. Once you know which failure you have, you can tell which story is true.

The five ways an epoxy floor fails

Two of the five, yellowing and chalking, only change how a floor looks. The floor is still doing its job. Two more, peeling and hot-tire pickup, are bond problems, and you can see where they stop. The last one, blistering, starts in the slab and only shows up in the coating. It’s the one failure on the list where a straight recoat is likely to fail the same way.

Why is my epoxy floor peeling?

Pick up a loose piece and turn it over. That’s the whole test, and it takes ten seconds.

A clean back, with no concrete on it. The coating never stuck to the slab at all. The cause is almost always the prep. Someone used acid, a solvent wipe, or a pressure wash where a diamond grind belonged. Acid etching leaves a texture of about CSP 1 to CSP 2. CSP is the roughness scale you’ll see on bids, and resin coatings need about CSP 3, a medium texture with peaks and valleys for the resin to grip. Texture alone isn’t the whole job, though. The grind also opens the pores, so the primer can soak in. To check, a good crew sets water drops on the ground floor in three spots per 1,000 square feet. Each drop should darken the concrete and soak in within 60 seconds. If a drop beads up, the surface is still closed, whatever the texture looks like, and the grinding keeps going. The surface preparation guide explains the CSP scale and how a crew checks it. This kind of peeling most often starts at a door sill or in a strip of sun. That’s where the floor heats and cools the most. Then it works its way in.

A gray skin of concrete stuck to the back. The bond held, and the top of the slab tore off instead. That means the slab surface was weak, or no one ground off the laitance. Laitance is the soft, dusty film that rises to the top of a slab while it’s being finished. This is a different problem, and a costlier one. The next coating needs a rougher grind to reach sound concrete, and a bad case needs a repair pass before anything else goes down.

Same symptom, but a different cause, a different job, and a different number on the bid. Anyone who quotes a peeling floor without turning a piece over is guessing.

Why do blisters form under a floor that looked fine?

Blistering comes from osmosis. Once you see how it works, the fix makes sense.

Concrete is full of tiny pores. Water vapor moves up through a slab all the time, wherever the ground under it is wetter than the air above. On a bare floor, that vapor just leaves, and no one notices. Put a sealed film over it, and the vapor collects at the bond line instead. That trapped water picks up salts from the slab. Salty water under the film pulls in more water through the concrete behind it, because water moves toward the saltier side. The pressure builds, the film lifts into a dome, and the dome grows.

That’s why a cut-open blister often holds water and leaves a white salt ring on the concrete. It’s also why a floor can look perfect for a year and then break out in blisters after one damp spring.

The vapor comes from the slab, not from a bad coat or a careless crew. So the warning signs matter more than the product. We check every slab. We run a moisture test when the check shows a warning sign, or when you ask for one. Warning signs include a slab poured before 1990, a slab with no vapor barrier you can see at the edge, or a floor that has blistered once already. The tests are ASTM F1869, a calcium chloride test, and ASTM F2170, a probe set into the slab. They answer the question before anyone picks a coating. Most makers set the limit at about 3 pounds per 1,000 square feet per 24 hours on the calcium chloride test. On the probe, it’s 80 percent humidity. Above that, the coating needs a moisture-blocking primer, or it shouldn’t go down at all.

There’s a free version you can run first, and the concrete moisture test walks you through it. It won’t give you a number, but it will tell you whether you need one. If your installer has already blamed moisture, check how their test was run before you accept it.

How do you repair epoxy blistering?

To repair epoxy blistering, deal with the water first. Start by taking the blistered coat off, back to sound film or bare concrete. Next, the tests above settle the moisture question. If the reading runs over the limit, a moisture barrier coat goes down before anything else. RESA (Resin-Encapsulated Seamless Aggregate) is our name for our full-broadcast resinous floor coatings. On our RESA Flake floors, that barrier is its own line on the quote at $1.80 a square foot. We call for the barrier when the calcium chloride test reads over 3 pounds. Only then does the new coat go on top.

What causes hot-tire pickup?

After a summer drive, tires reach 150 to 200 degrees F, and they hold that heat for a while after you park. Two things have to be true for the tire to win.

The first is the chemistry. Standard epoxy gets softer as it warms up. A soft film and a hot rubber tire stick together for a while. When you pull out the next morning, the bond between the coating and the concrete turns out to be the weaker one. What’s left is tire-shaped patches of bare slab, right where the car sits.

The second is thickness. A store-bought water-based epoxy goes on about 2 to 3 mils thick, which is about two sheets of printer paper. A built system runs 18 to 22 mils across several coats. Thin film over a smooth slab is the mix that fails. The same chemistry at a real thickness over a ground slab holds up far longer. The garage floor lifespan breakdown shows the thickness and lifespan ranges side by side.

Hot-tire pickup only happens where a car parks, so the damage has a clear edge. That’s good news for the size of the job, and it’s why this failure often ends as a repair instead of a replacement.

Why does epoxy turn amber near the door?

Standard epoxy resins are aromatic compounds. UV light breaks bonds in their chains and frees chromophores, which are tiny bits that carry color. They shift a clear or light film toward yellow or brass. In direct sun, the change shows in 6 to 18 months. It can’t be undone, and no cleaner or wax will take it out.

Where it happens gives it away. Yellowing is worst where daylight lands: inside an overhead door, under a window, or along a south or west wall. Walk twenty feet into the building, and the same coating is still the color it started. If the color change follows the light and not the foot traffic, it’s UV and nothing else.

This is the cheapest failure on the page. Nothing has come loose, and nothing is lifting. If the film is sound, the fix is a light sanding and a recoat with a UV-Stable topcoat. That topcoat is an aliphatic kind, like polyaspartic or urethane, that takes UV without breaking down. Be careful with a bid that promises a UV-stable pure epoxy. The polyaspartic and epoxy comparison explains why that claim most often means an epoxy base with a different topcoat over it. That’s a fine system when the bid says so plainly, and a bad one when it hides it.

What is the chalky film on the surface?

Run your hand across it. If you come back with a light powder, and the coating under it is still stuck down, that’s chalking. The resin at the very top of the film wears away under sun and weather, and it leaves loose color and filler behind. Chalking shows up in the same sunny spots as yellowing, and often on the same floor.

It’s easy to mix chalking up with a dusting slab, but the two need different fixes. Chalking is a coating breaking down at the top. A dusting slab is bare concrete with a weak top layer, which comes from the pour and the cure and has nothing to do with any coating. The test is whether there’s a film at all. Scratch through the powder with a key. A solid colored layer under it means a coating. Gray concrete under it means you’re looking at the slab itself.

If the film is still bonded, chalking is upkeep. A light pass with 60 to 80 grit knocks the powder off and opens the surface. Then a fresh topcoat goes on, and you’re done. Plan on that every 5 to 10 years on a bay that gets sun. If the film is chalking and lifting at the same time, the lifting is the problem to look into.

Two honest fixes for a failed floor

Everything above comes down to one of two answers, and the size of the damage picks which one.

If the coating is stuck down everywhere but one spot, you have a small problem and a small fix. Cut the failure back to sound coating, taper the edge, and repair that zone. On a garage-sized patch, that’s a weekend of work, and the peeling epoxy repair procedure walks you through it step by step. A spot repair will always look like a patch. That’s a real cost, and it’s worth knowing before you start. But it costs less than redoing a floor that mostly works.

If the failure comes from moisture, or most of the floor has let go, the answer is harder. Grinding a failed coating off a big floor is slow, loud, dusty work, and it takes real machines. Putting a second store-bought product over the same slab that killed the first one is the surest way to pay twice. On a floor in that shape, you can strip it to bare concrete and leave it bare, swept and sealed. That’s a real finished floor, and it costs far less than a second failure would.

Doing your own floor is fine. Two mistakes cause nearly all of these failures: skipping the grind, and putting a film over a slab that’s still pushing up water. Get those two right, and a coating lasts. Get either one wrong, and the name on the bucket won’t save it.

Read the failure first. The floor is telling you which one you have, and it’s the only voice in the room that isn’t selling you something.