procurement · testing
Running a sealant sample trial that means something
Updated 2026-11-10
A sample that “seemed fine” is not a trial. A trial has a control, a defined substrate, recorded cure conditions and a failure test that tells you how the bond breaks.
The minimum protocol
- Define the substrate. The actual panel, coating or glass, from the actual supplier.
- Prepare it as production will. Same cleaner, same abrasion, same primer decision.
- Apply at the production bead size. A thin laboratory smear tests nothing.
- Cure at the production temperature and humidity. Record both.
- Test to failure. Peel or shear, and classify cohesive vs interfacial.
- Run a control. The product you use today, same conditions.
Without the control, you know the sample bonded; you do not know whether it is better or worse than what you have.
Reading the failure
| Failure mode | Meaning |
|---|---|
| Cohesive (adhesive tears) | Bond stronger than the sealant; product is doing its job |
| Interfacial, clean substrate | Preparation or compatibility problem |
| Mixed | Partial adhesion; investigate surface first |
| Foamed or crumbly | Moisture or storage problem |
What a sample cannot tell you
- Long-term durability. That needs weathering or cyclic testing, not a week in a workshop.
- Production consistency. One cartridge is not a batch.
- Cost per metre. That comes from pack format and waste.
How many coupons, and on what
One coupon proves the adhesive bonded once. A trial that predicts production needs a small matrix:
3 substrates × 3 coupons × 2 products (sample + control) = 18 couponsThat is one day of work and two cartridges. It answers three questions a single coupon cannot:
- Does the result repeat? Three coupons on the same substrate show the spread.
- Does it transfer? Three substrates show whether the claim is general or specific.
- Is it better? The control gives the comparison its meaning.
Where 18 coupons is too many, the minimum is three coupons on the actual production substrate with the control run alongside. Anything less is a demonstration, not a trial.
The record sheet
A trial without a record is an anecdote. Six fields make it evidence:
| Field | Example |
|---|---|
| Batch number | As printed on the cartridge |
| Substrate and supplier | Coil reference, coating type |
| Preparation | Cleaner, abrasion, primer, flash-off time |
| Cure conditions | Temperature and RH, recorded at start and end |
| Cure time before test | Hours, and bead section |
| Failure classification | % cohesive vs interfacial, with a photo |
The photo matters as much as the number. Two people can look at the same failure surface and classify it differently; a photograph against a ruler settles it.
What changes between a sample and a production batch
A sample cartridge and a production pallet are made on different days, sometimes on different lines. The variables that move:
| Variable | Why it matters |
|---|---|
| Raw material lot | Pre-polymer viscosity shifts batch to batch |
| Line speed | Faster filling can trap air |
| Filler dispersion | Affects extrusion rate and sag |
| Cure age at shipment | A fresh batch behaves differently from one three months old |
This is why the sample batch number should be written into the order, and why the first production batch is tested against the sample rather than against the catalogue. The catalogue describes the design; the sample describes what the factory can actually make.
What this means for buyers
- Always run a control alongside the sample.
- Record temperature, humidity and cure time — otherwise the result is not repeatable.
- Test to failure, not to “looks good”.
- Keep the sample batch number and compare it with the first production batch.