Repair, Recoat or Replace an Industrial Floor: A Decision Matrix

The rule that governs every option

A worn industrial floor presents the same three options everywhere: repair the damaged parts, recoat the whole thing, or take it off and start again. The choice looks like a budget question and is usually argued as one, but it is not. It is a bond question. You can only recoat a floor whose existing system is still soundly bonded to the concrete. Everything else follows from that single test.

The reason is mechanical. A coating carries load in shear through the bond line into the slab. A new topcoat applied over an existing system inherits that bond line unchanged — it adds mass and stiffness above a plane that has already failed, or is in the process of failing. Where the old system is drummy, blistered, or sitting on a dusting substrate, the new coat repairs none of it. It conceals it. The floor looks correct at handover and delaminates in sheets a few months later, at which point the client has paid twice and lost the shutdown window twice.

So the useful question is never "should we recoat or replace?" asked in the abstract. It is "what proportion of this floor is still soundly bonded, where is the failure plane, and is the slab dry enough to accept anything at all?" Answer those three and the specification largely chooses itself. This article sets out the diagnostics, the three strategies, and a matrix that maps observed condition onto the right one. Money is dealt with separately in our article on industrial floor coating repair cost.

The three strategies, and what each one assumes

Localised repair assumes the floor as a whole is sound and the damage is bounded. Impact craters at a rack leg, a few cracks in one aisle, one bay of accelerated wear where a turning point has scuffed through to primer. The work is to saw-cut or grind out the damaged area back to a sound perimeter, reinstate the substrate, re-prime and patch-coat. Sparco Epoxy Mortar — a three-pack of base, hardener and aggregate, non-shrink, steel-trowel applied and touch dry in about four hours at 30 °C — is a conventional choice for reinstating the removed section before coating over it. Localised repair is the cheapest option and the shortest downtime, and it will always leave a visible patch. Anyone who tells you otherwise is selling a recoat.

Recoat, or overcoat, assumes the existing build-up is bonded and structurally sound but has run out of service life at the surface, or was never right for the process the building now runs. The sequence is: mechanically abrade the whole area to create a key, spot-repair the defects you find while abrading, then apply a new topcoat or a full new build-up over the prepared surface. Two conditions must be satisfied before this is defensible. The first is adhesion: the existing system has to pass a pull-off test against the acceptance criterion set out in the project specification or the coating manufacturer's literature. The second is compatibility: the new chemistry must be confirmed as suitable over the old, which is a technical review, not an assumption.

Full removal and replacement is what widespread debonding, osmotic blistering, a dusting or weak slab, moisture above the coating's limit, embedded contamination, or a change of use the current build-up cannot serve all point to. The old system comes off to sound concrete, the substrate is re-prepared mechanically, retested, and the floor is re-specified from the slab up. Where the slab itself is the problem — weak surface concrete, oil that has migrated into the pore structure, no functioning damp-proof membrane beneath a Singapore slab-on-grade — replacement is the only route that addresses the cause rather than the symptom. A self-smoothing solvent-free system such as Sparcofloor SL 200, or a roller-applied high-solids coating such as Sparcofloor #102, may then be specified over a properly re-primed substrate, depending on traffic, chemical exposure and downtime.

• Localised repair: sound floor, bounded damage. Cheapest, fastest, visibly patched.
• Recoat: bonded and sound, but worn or wrongly specified. Requires an adhesion test and a compatibility check.
• Replacement: the bond, the slab or the moisture regime has failed. Nothing applied on top corrects any of those.
• A chemical key alone is rarely sufficient on a fully cured epoxy — a mechanical key is the norm.
• Any of the three may be phased bay by bay where the facility cannot surrender a single continuous shutdown window.

The diagnostics that drive the decision

Start with a survey. Walk the floor, map the defects on a drawing, and record what kind of defect each one is. Cratering is not the same as blistering, and neither is the same as a coating that is simply worn thin. Then sound the floor. A hammer or a chain drag across the surface will find drummy areas — a hollow return where the coating has released from the slab. Mark them and measure them. The percentage of area that returns a drummy response is the single most decisive figure in the survey, because it converts an aesthetic complaint into an engineering condition.

Follow the sounding with pull-off adhesion testing to ASTM D7234, Standard Test Method for Pull-Off Adhesion Strength of Coatings on Concrete Using Portable Pull-Off Adhesion Testers. This is the concrete method — not ASTM D4541, which is the metal-substrate method and should not be cited for a coating on a slab. Where the coating is thicker than roughly 20 mils, it must be scored down to the concrete before the dolly is attached, or the result reports the tensile strength of the coating rather than the strength of the bond. Test in every condition zone, not only in the good areas. We deliberately publish no minimum acceptable pull-off value: the acceptance criterion belongs to the project specification or the coating manufacturer, and quoting a figure out of that context is how a floor gets condemned or approved for the wrong reason. For context, Sparco's TDS for Sparco Epoxy Bonding Primer #100 calls for a minimum prepared-substrate pull-off strength of 1.5 N/mm² — a requirement for the prepared concrete, not a pass mark for an aged coating.

Then read the failure plane. When the dolly comes away, look at what is stuck to it. Concrete on the back of the dolly means the bond outperformed the slab, and the substrate is the weak element. A clean release at the coating-to-concrete interface means adhesive failure at the bond line. Separation between two coats means an intercoat failure, which changes the diagnosis again. Finally, test slab moisture: ASTM F2170 measures internal relative humidity using in-situ probes, and ASTM F1869 measures moisture vapour emission rate using anhydrous calcium chloride. Both are test methods; the acceptance limit is set by the coating manufacturer or the project specification, not by the standard. Our moisture-testing article covers procedure, probe depths and conditioning in full. In a Singapore slab-on-grade with no functioning damp-proof membrane, this test decides more floors than the adhesion test does.

Compatibility, cure state and the shutdown window

An aged epoxy is fully cured and largely inert at the surface. There is very little left for a new coat to react with, which is why a solvent wipe or a chemical etch on its own is rarely sufficient to guarantee intercoat adhesion. The answer is a mechanical key: abrade the whole surface so the new coat has profile to grip. That is a different operation from sanding a paint film smooth, and it produces a different result. Where a coating is being applied to exposed concrete rather than to an existing film, the requirement is mechanical preparation by ball blasting, milling or diamond grinding, with weak concrete removed rather than coated over. Our surface preparation article covers profile targets and ICRI Guideline No. 310.2R-2013 in detail.

Compatibility between old and new chemistry is the second gate. A polyurethane or polyaspartic topcoat over an epoxy body coat is a common and well-established build-up — the epoxy provides film build and adhesion to the slab, the PU provides abrasion resistance and colour stability. That does not make every PU compatible with every aged epoxy, and it certainly does not make an unidentified legacy coating a safe base. Where the existing system cannot be established from records, a trial area with pull-off testing after cure is worth more than any assumption, and the overcoat should be confirmed with its manufacturer. Sparco Epoxy Bonding Primer #100 is a conventional re-priming choice where the coating has been taken back to concrete: its TDS caps permissible concrete substrate moisture content at 5% and calls for a minimum prepared-substrate pull-off strength of 1.5 N/mm².

Downtime is not a footnote to the decision, it is part of it. A full removal and replacement needs a continuous shutdown window measured in days, with grinding or blasting noise, dust containment, and a cure period before traffic returns. A 24/7 Singapore facility may simply not have that window, and saying so is not an excuse — it is a real constraint that changes how the work is planned. The honest response is to phase the work bay by bay, taking a defined area out of service at a time, accepting the cold joints between phases, and sequencing around production. What is not legitimate is allowing a short window to downgrade a floor the survey has already classified as a replacement into a recoat. That trade buys months and loses the window twice.

Common mistakes

Recoating over blisters. A blister is a symptom of pressure beneath the film — commonly osmotic, driven by moisture and soluble material below the coating. Grinding the blisters flat and coating over the area re-establishes exactly the conditions that formed them, now under a thicker and stiffer membrane. The blisters return, usually in the same places.

Sanding rather than mechanically abrading. Sanding smooths a coating. It does not create the profile a new coat needs to key into a fully cured epoxy. The two operations produce visually similar floors and materially different bond strengths, and the difference does not declare itself for a year.

Assuming a new topcoat cures a substrate problem. It does not. A wet slab stays wet, a dusting slab stays friable, and contamination stays in the pore structure. Coatings are surface systems; substrate defects are substrate work. Where the slab is the problem, the slab is the scope.

Choosing replacement when a phased repair would serve. The opposite error is real too. A floor with isolated impact damage and a bond that is sound across the great majority of its area does not need to be removed because it looks tired. Full replacement of a serviceable floor is a large and avoidable outlay of both money and production time.

Deciding before testing. This is the expensive one. Every strategy above is defensible on the right floor and indefensible on the wrong one, and the only thing separating the two is a survey, a sounding, an adhesion test and a moisture test. Specifying a system before those results exist is a guess wearing a specification's clothes.

Checklist: what to have in hand before you decide

The decision matrix is only as good as the evidence fed into it. Before a specification is written — and certainly before a contractor is asked to price the work — the following should exist as a written record, ideally with photographs and a marked-up floor plan. This is also the pack that allows two contractors to quote the same scope rather than two different guesses at it.

If an item on this list is missing, the correct next step is to obtain it, not to proceed. A survey costs a fraction of a floor, and it is the only part of the process capable of saving the whole of it. Where the facility cannot be taken out of service to survey, the survey can usually be phased in the same way the work would be.

• A marked-up floor plan showing every defect, classified by type: impact, crack, wear-through, blister, delamination, dusting, contamination.
• The proportion of total area returning a drummy response to hammer or chain drag, with the drummy zones located on the plan.
• Pull-off adhesion results to ASTM D7234, taken in each condition zone, with coatings over about 20 mils scored down to the concrete before testing.
• A description of the failure plane at each pull-off: concrete on the dolly, clean bond-line release, or intercoat separation.
• Slab moisture results to ASTM F2170 or ASTM F1869, and the acceptance limit taken from the coating manufacturer or the project specification.
• Identification of the existing system where records exist, and a written compatibility confirmation for any proposed overcoat.
• A statement of what the floor is being asked to do next: traffic, chemical exposure, thermal exposure, washdown regime, hygiene requirement.
• The realistic shutdown window, and whether phased bay-by-bay working is available as an alternative.

Frequently asked questions