UV adhesive in flexible lamination — delamination doesn't show up today, it shows up at your customer's plant.

The problem

Three variables that have to line up — and a defect that shows up weeks later.

A sound film-to-film bond with UV adhesive is the intersection of three realities that most plants do not cross-check in-line, only after the fact:

1. The adhesive viscosity at the dosing unit — it depends on temperature, on the supplier's batch and on the condition of the applicator. If it rises, the actual coat weight changes even though the setting has not been touched.
2. The coat weight deposited on the film — the product target. Sometimes it is measured by weight difference, sometimes assumed from the recipe. Rarely is it measured in real time.
3. The effective UV dose — the mJ/cm² the adhesive actually receives. It depends on lamp power, film speed, distance and bulb age. It is usually monitored at setup, not on every roll.

When all three are inside the window, the bond holds. When one drifts, the initial tack can pass the plant check and the laminate delaminate weeks later — when it enters the customer's cold room, goes through the packing cycle or gets stressed on the end-of-line equipment. Or worse: the roll comes out slightly yellowed from UV overdose and the customer rejects it as a cosmetic defect. The damage is to the brand, not just the roll.

How it fits the IRIS system

iLEAN does not replace your control lab — it cross-checks the three curves you already have before rewinding.

The delamination problem is not a lack of measurement: it is that viscosity, coat weight and UV dose are never cross-checked in the same second against the product window. iLEAN acts as the putty that seals that crack, without asking you to change your laminator, your adhesive or your lamp.

Edge sees the roll and measures the dose. Connect reads the viscosity and the product recipe. The agent cross-checks the three curves against the target window and holds the roll if something does not add up. The person signs — the line does not restart on its own.

The three iLEAN pieces applied to UV adhesive control in flexible lamination:

• iLEAN Edge — a vision (CNN) terminal over the film at the UV lamp exit, reading color (yellowing), continuity of the adhesive layer and, with the right sensor, in-line UV dose. It works without a network: if the plant loses WiFi, Edge keeps reading and holding rolls before rewinding.
• iLEAN Connect — captures the adhesive viscosity from the in-line meter, the dosing-unit temperature and the product recipe whether it comes from the ERP, the MES or an Excel sheet where R&D logs the % solids of the supplier's latest batch. And it also captures the supplier's email announcing a batch change, at second zero, without anyone having to forward it.
• Agent — cross-checks viscosity, estimated coat weight, UV dose and product recipe against the signed window. If the cross-check falls outside it, it holds the roll and alerts the quality manager through whatever channel they use. At closeout, it leaves a per-roll dossier with the three curves for the BRC auditor or for the customer's claim.

See the full IRIS architecture →

Before and after

Isolated control vs. cross-checked control with iLEAN

Aspect	Classic control	With iLEAN Edge + Connect + Agent
Adhesive viscosity	Spot measurement, not always cross-checked	Continuous, against the product window
Actual coat weight	By weight difference or assumed	Estimated in-line from vision and dose
UV dose	Measured at setup, assumed afterward	Monitored every meter, alarm if it drops
Yellowing detection	Operator's eye — or the customer's	CNN camera before rewinding
Roll hold	After the fact, in the lab	In-line, before it leaves the plant
Dossier for the BRC auditor	Rebuilt by hand	Curves + recipe + signature, per roll

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Flexible packaging plant, one or several film-to-film laminators running UV adhesive, multi-product with frequent recipe and substrate changes.
• Edge pilot on one laminator (CNN camera at the UV lamp exit + integration with the viscometer and the lamp control). First value expected within a few weeks.
• Expected reduction in delamination / yellowing claims ≥ 30% against the documented frequency of recent months.
• Indicative payback between 4 and 9 months, depending on the average cost of a claim (product recovered, reverse logistics, brand damage) and the volume of multi-customer rolls.
• The hard lever is a single delamination claim avoided: over the medium term the brand cost outweighs the immediate hard cost. One claim pays for the pilot.

And the quality manager's reasonable doubt

“What if the AI holds a good roll and stops the line for no reason?” — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI merely recontextualizes a piece of data from one system to another (reading color, coat weight and dose and comparing them against the product window), the best models brought error below 1.5% ^[1]. And even so, what is critical is never decided alone: iLEAN holds and the person signs whether the roll goes to the customer, to rework or to scrap. The three safety rings exist precisely for this.

^[1] OpenAI paper “Why Language Models Hallucinate”, 2025 — on the reliability of AI in anchored tasks.