Application

DFT measurement that counts: applying the 85-15 rule on site

Film thickness looks like the least controversial measurement there is: press the gauge, read the number, compare with the specification. In practice it generates more disputes than surface preparation, because "what counts as a pass" has five or six different definitions.

Three are common: judge by average, apply the 80/20 rule, or require every point above a floor. All three can be described as testing to standard, and the conclusions can be opposite.

1. Two families of rules: average and point-by-point

SSPC-PA 2 works on average plus a floor: take a number of readings over a defined area, require the average to be at or above the specified thickness, no single reading below 80% of specified, and no more than 20% of readings below specified. This is the most widely used basis internationally.

ISO 19840 follows similar logic in different words: the average governs acceptance, a minimum individual reading is defined (usually 80% of nominal), and the number and layout of readings are specified. The two standards are frequently mixed in practice, so a contract saying "to ISO 19840" and one saying "to SSPC-PA 2" differ in execution detail.

A stricter basis also exists: every point at or above 90% of nominal, or even every point at or above nominal. It appears in offshore and some foreign-invested projects, at the cost of significantly more over-thickness and rework.

The point is not which is more scientific but that one basis is fixed before work starts and the applicator knows it. Changing the basis afterwards is the surest route to a settlement dispute.

2. Where to measure: both count and location matter

SSPC-PA 2 sets the number of readings by area, roughly one set of three readings per 100 square metres, with more required over larger areas. ISO 19840 gives a similar density requirement. This is not formalism: blasting and spraying produce uneven thickness, and too few readings make the average meaningless.

Location is easier to get wrong than count. The high-risk places are edges, welds, the backs of bolts, areas shadowed by brackets, and the start and finish of a spray pass where gun distance is unstable. Thickness is usually low there, but measure only the middle of flat plate and the data looks unrealistically good.

A practical compromise: sample flat areas at the standard density at random, and add dedicated readings at known risk points, recorded separately. The report then has two columns — average and pass rate for the main area, individual values for risk areas. That satisfies the standard without missing the real weak spots.

3. Substrate and surface condition mislead the gauge

A magnetic gauge reads change in magnetic flux, so substrate thickness, material and surface roughness all affect the reading. Blasted surfaces are the worst case: the probe sitting on a peak reads high, in a valley reads low, and repeated readings on one plate can vary by ten microns or more.

Correct practice is to calibrate on a smooth coupon of the same material and apply a roughness correction for blasted surfaces, usually subtracting a value related to profile. Many sites use the default calibration on blasted steel, the readings run systematically high, and the "pass" on the report is unreliable.

Curvature matters too. Below a certain diameter, around 25 millimetres, the probe cannot sit flat and readings distort, so a curved probe or a coupon comparison is needed.

Another case is soft coating. On a film that has not fully cured, the probe sinks in and reads low. Epoxy products cure slowly at low temperature, so measuring the next day often gives low values that normalise once cure is complete. Not knowing this leads to pointless additional coats.

4. Over-thickness is a problem too

Everyone watches for thin spots at acceptance and almost nobody watches for thick ones. But excess thickness has consequences: higher internal stress and cracking, particularly with inorganic zinc and high-build epoxy; trapped solvent and long-term blistering; and wasted material, which shows up directly in cost.

Some standards set a maximum, for example not more than two or three times nominal. Technically the greater concern is local excess — welds and edges accumulate paint when brushed, and a 200 micron thick edge is more dangerous than a uniformly thick field because stress concentrates there.

A report should therefore show the maximum and the distribution, not only the average. Average alone hides local excess.

5. Settle the dispute in advance

I would write three things into the specification: the acceptance basis (which standard and which rule), the reading density and locations including additional readings at risk areas, and the calibration method including whether roughness correction applies.

Add one more: retest and arbitration. For example, where the initial result fails, allow double readings around the same location; where disagreement remains, a third party retests to the same standard at the cost of the responsible party. Spelling that out reduces the shouting considerably.

One last observation from experience: fewer than thirty percent of thickness disputes are genuinely technical. Seventy percent are rules that were never fixed. Fixing the rules costs one page; not fixing them costs a rework and an argument.

Need selection advice?

Tell us your substrate and service conditions

One-to-one coating consultancy: system design, quantity take-off and quotation.

Contact us
WhatsApp