Pull-Off Adhesion Testing on Epoxy Floors: How to Read the Results

Two adhesion reports can show the same MPa number and mean completely different things. The key lies in the failure mode: where the test specimen breaks tells you which bond is the weakest link. This article explains how to read pull-off adhesion test results on epoxy floors — as a general guide to reading results. It does not publish any MPa value or failure mode for Kadier, because the figures for each individual coating system are only found in that specific product’s TDS and its accompanying test report; every general reference range below carries the same note: a general reference range only — the specific product’s TDS decides.

Illustration: a laboratory worker handling test instruments. Source: Pexels, by Pavel Danilyuk. Illustration only — not a Kadier pull-off test session or project.

1. What does the pull-off test measure?

The pull-off method (ASTM D4541; the equivalent standard commonly cited alongside it is ISO 4624) measures adhesion strength by gluing a cylindrical dolly to the surface, waiting for the adhesive to cure, cutting around the dolly down to the substrate to isolate the test area, then pulling perpendicular to the surface at a steady loading rate until failure. Strength is calculated as the failure force divided by the dolly area, in MPa.

The crucial point: failure always occurs at the weakest plane of the system — it may be the concrete substrate, a coating layer, the interface between layers, or the dolly adhesive itself. So an MPa number never stands alone: it only means something together with the failure mode, dolly size, loading rate, substrate condition and number of test points.

On the domestic standards framework: this article only names TCVN 12574-2018 in the form “under the TCVN 12574-2018 framework (the full text should be checked before citing any clause)”. It cites no clause, article or MPa threshold from TCVN 12574-2018, because the full text of the standard was not checked against during the writing of this article.

2. Read the failure mode first, the MPa number second

A proper report does not just record a bare MPa number; it records the percentage of area for each failure mode (for example, a notation such as “B-100%” in ASTM D4541 means 100% cohesive failure within layer B). The four main modes:

Failure mode	What you see	What it means
Adhesive failure	The coating peels cleanly off the substrate or cleanly off the layer below, with a smooth separation surface	The bond at the interface is weak: possibly inadequate substrate preparation, a contaminated substrate, an undercoat over-cured before the next coat, or incompatible layers.
Cohesive failure	The break lies inside one coating layer, which itself cracks/breaks	That layer itself is weaker than its bonds: check its thickness, degree of cure and brittleness.
Substrate failure	Concrete breaks along with it, with concrete fragments stuck to the dolly	Usually a good sign — the coating’s adhesion is stronger than the concrete’s own tensile strength — but the MPa number is then the substrate’s limit, not the coating’s. Do not record that number as the “coating’s adhesion strength”.
Glue failure	The dolly detaches from the adhesive, or the adhesive itself breaks before the coating system fails	The result is only a lower bound, is usually treated as invalid, and needs retesting with a stronger adhesive or method. A failed adhesive proves nothing about whether the coating is weak or strong.

3. Why a high MPa is not necessarily good, and a low MPa is not necessarily a failure?

Compare two hypothetical cases (illustrative examples, not Kadier figures): case A fails in the substrate at a moderate MPa — the coating bonds more strongly than the concrete itself; case B fails adhesively at a higher MPa but the coating peels cleanly off the substrate. Case A is usually more trustworthy, even though its number is lower, because the weakest plane lies in the substrate, not in the coating–substrate bond.

Conversely, a high MPa number accompanied by adhesive failure is still a bad signal: it shows the bond to the substrate failed before the other layers could perform. Comparison is only valid when reports share the same dolly size, loading rate, substrate condition and failure mode. Not every application has a universal pass/fail value — the expected value is set by the parties, the contract and the TDS for that project.

4. Five errors that commonly distort pull-off results

1. Off-axis dolly during pulling: the force is no longer purely perpendicular, so the result deviates from true adhesion.
2. Ring cut not reaching the substrate: the test area is not separated from the surrounding coating, so the pull is partly “carried” by it.
3. Adhesive not fully cured: easily leads to glue failure, leaving the result as a lower bound only.
4. Damp or dirty substrate, or coating not fully cured at the time of testing: the result reflects the test conditions, not a stabilised coating system.
5. Too few test points: a single point does not represent a whole floor. As a laboratory reference practice, typically 3–5 valid points are needed for each location — this is reference practice, not a Kadier threshold, and the specific number of points remains set by the contract and the acceptance method of each project.

5. Pull-off on a real floor: agree before testing

Pull-off is a locally destructive test: after pulling, the floor is left with a dolly mark that needs repair. So before testing, the parties should agree in writing: where the test locations are, how many points, which dolly size, when to test (after the coating has fully cured), and how the test marks will be repaired and who is responsible for the repair. Do not promise “testing without leaving a mark” — that is not true to the nature of the method.

Pull-off is also only one item in acceptance, not the whole of acceptance. A floor also needs film thickness checks, appearance checks, substrate moisture checks and a handover file — see the acceptance and warranty page and the article pull-off adhesion testing. For abrasion, see What Do Shore D Hardness and Taber Abrasion Tell You?.

Frequently asked questions

When you receive a pull-off report, what should you read first?

Read the failure mode and the percentage for each mode first, then the MPa number. If a report gives only an MPa number without the failure mode, dolly size and substrate condition, it is not enough to draw a conclusion.

If the dolly adhesive fails, can the result still be used?

Only as a lower bound: the system withstood at least that much without failing. Such a case is usually treated as invalid and needs retesting with a stronger adhesive or method.

Is substrate failure a good result?

Usually a good sign, because the coating bonds more strongly than the concrete itself, but the MPa number is then the substrate’s limit, not the coating’s adhesion strength — it should not be quoted as a coating parameter.

Why does this article not state Kadier’s passing MPa threshold?

Because a project’s passing threshold is set by the contract and the TDS of the chosen coating system, together with the method and failure mode. This article is a general guide to reading results, not a substitute for those documents.

Contact

Before construction, agree the pull-off test method and locations with Kadier via the quote page or the contact page (hotline/Zalo 0899355489, WeChat wxid_95o54joujyrr12), together with information on the existing substrate and the area’s function, so you can be advised on suitable acceptance items. Also refer to the RollCoat, LevelFlow, StatGuard and AquaBase systems.