Three Pull-Offs and a Cold Adhesive

Test conditions first - the whole set is suspect. The adhesive was cured 12 hours at 41°F against a 24-hour cure at 70°F requirement. Adhesive cure at 41°F is dramatically slower, so the glue was almost certainly under-strength. Any low result may be measuring the adhesive, not the lining.

Individual evaluation:

• Dolly 1 (610 psi, adhesive at dolly/coating interface): this is a glue-line failure - the adhesive let go of the coating, not the coating letting go of steel. The result is invalid and must be discarded, not recorded as a 610 psi coating failure. It is also not automatically a pass; it is simply a null result.
• Dolly 2 (820 psi, 60% cohesive in the epoxy): above 750 psi and the failure is within the coating, meaning the bond to steel exceeded 820 psi. Valid pass - provided the adhesive was sound.
• Dolly 3 (1150 psi, 100% cohesive): valid pass.

Conclusion: two results pass on paper, but because the test temperature and adhesive cure were outside the method's requirements, the whole set should be re-run at a controlled temperature with proper adhesive cure before formal acceptance. Do not sign off on a set that includes one invalid pull and uncontrolled conditions.

What cohesive failure means: The break occurred inside the epoxy film, so the epoxy-to-steel bond is stronger than the epoxy itself. The coating is the weak link, and at 820 and 1150 psi it is performing above the 750 psi requirement. Cohesive failure in a fully cured, properly applied 100% solids epoxy is a normal and acceptable outcome.

What would change the recommendation:

• Adhesive failure at the coating/steel interface at low load - this is a genuine red flag. It points to surface preparation problems: contamination, oil, residual salts, moisture, amine blush, or coating applied to a cold or damp substrate. In this job the 41°F surface is the obvious suspect.
• Low values that rise after full cure would confirm the lining was simply undercured at test time and needs time, not rework.
• Adhesive failure at the coating/coating interface would suggest intercoat contamination or exceeding a recoat window.
• A wide scatter across the tank (e.g., 1150 psi in one area, 300 psi in another) means the survey is not representative and more pulls are needed in the low area to define the extent of repair.

Also verify method details that invalidate results if missed: dollies abraded and cleaned before gluing, the coating scored through to steel around each dolly, the dolly face perpendicular to the pull, and the pull rate per the method.

Reporting:

• Record for each dolly: location on a tank map, dolly diameter, peak load in psi, failure mode (adhesive vs. cohesive), percent cohesive/adhesive, and the failure plane described precisely (dolly/adhesive, adhesive/coating, coating/steel, cohesive within coating).
• Record test conditions: surface temperature, air temperature, RH, adhesive type, cure time and temperature, gauge model and calibration date, and the technician.
• Report Dolly 1 explicitly as an invalid test due to glue-line failure, not as a coating failure - and state that re-testing was performed.
• Reference the governing standard/specification clause and state the acceptance conclusion and any owner concession.

Repair:

• Every pull-off site is a destructive test: the dolly area and scoring ring must be repaired. Abrade the damaged area, solvent-clean, fill and build back to 20–25 mils with the same product, respecting the recoat window.
• Holiday-test each repaired patch (ASTM D5162 / NACE SP0188 practice) and re-verify DFT - a small patch in a lining is exactly where a pinhole hides.
• Perform the re-test pulls at the specified frequency in the required locations, at controlled temperature, and only release the tank after both the DFT survey and holiday survey are closed out and the documentation package is signed.