Is 5 MPa a passing pull-off result? Look at where it broke first
Pull-off testing is the easiest item on a coating acceptance list to reduce to a single number. A report giving only an average, with no failure location, is almost useless — 5 MPa might be a good result or might mean the test was done wrong.
1. Learn the four failure planes
Type A: between adhesive and coating. That is a test problem, not a coating problem — poor bonding or incomplete cure, and the value is void. Type B: within the coating, cohesive failure, meaning the film itself is weak.
Type C: between coating and substrate, adhesive failure. This is the only one that measures adhesion. Type D: within the substrate, concrete or base failure, meaning substrate tensile strength is below the coating's adhesion — a high number here measures concrete, not paint.
Specifications usually require 'adhesive failure not below X MPa' and a report stating failure type and area percentage. A number without the type is not a test.
2. Reading the number
For epoxy systems on steel, 5 MPa or more is a common pass. Zinc-rich primers run lower, 3–5 MPa, because the zinc reduces resin continuity. On concrete, 1.5–2.5 MPa is already good, since surface tensile strength is intrinsically low.
A number far above expectation is also a warning: 8 MPa on concrete usually means failure occurred inside the coating rather than at the interface, so adhesion was never measured.
3. Four common errors in the test itself
Insufficient abrasion so the adhesive cannot key. Inadequate adhesive cure, especially in the cold. Incomplete cutting — scoring the coating but not reaching the substrate, so the result is bulk tensile strength. And pulling too fast, where dynamic loading inflates the figure.
All four are avoidable with procedure. If a report omits failure type, test temperature and cut depth, it is not a rigorous report.
4. When it is mandatory
Whenever the design documents require it; on functional coatings such as fire protection, tank linings, potable water and ballast tanks; and in any blistering or detachment dispute, where pull-off is the most direct evidence.
Sampling is typically one set of three points per 500–1000 m². Points must include flat plate, edges and welds — welds usually give the lowest adhesion, so testing only flat areas yields an optimistic picture.