The lab confirms your basis weight complies — four hours after it stopped complying.

The problem

Spunbond basis weight moves before the lab finds out.

On a spunbond line, g/m² is the number that decides whether the roll is worth anything or not. In hygiene and medical applications, customer specifications demand tight tolerances (±5%) on a basis weight that depends on variables changing within the shift: the viscosity of the polypropylene in the new batch, the calender temperature after a recipe change, the extruder speed upstream.

The three realities almost nobody cross-references in the plant in time:

1. What g/m² is coming out right now — the web sensor (beta gauge/IR) measures it, but the reading lives on a screen few people watch.
2. Why it is moving — the cause is in the calender, the extruder or the polymer batch, data scattered across different systems.
3. Whether the current roll is usable or not — the lab confirms it once the roll is already wound, finished and about to be palletized.

The classic system (sampling every shift + SPC review) works 99% of the time. The remaining 1% is the roll the customer rejects, the batch that gets reworked or sold as second grade, and the claim that lands on the sales rep's desk.

How it fits the IRIS system

iLEAN does not replace the sensor — it adds an agent that cross-references signals and anticipates the drift.

The problem with basis weight is not a lack of measurement: it is measurement without context. Every sensor measures its own variable well and lives on its own island. iLEAN acts as the putty between the web, the calender, the extruder and the polymer batch, without asking you to change your MES, your sensors or the line.

Edge reads the sensors. Connect captures the polymer batch (spreadsheet, delivery note, ERP). The agent cross-references it with the line history and anticipates the drift — and proposes the adjustment. The person signs.

The iLEAN pieces applied to basis weight control:

• iLEAN Edge — a terminal over the line that captures in real time the g/m² from the web sensor, the calender temperature (thermocouple or a photo of the old display), the extruder speed and, optionally, a camera for visual inspection of the web. It works with no network.
• Connect — captures the incoming polymer batch whether it comes from the supplier's delivery note, from an ERP field or from the R&D manager's spreadsheet with the latest technical data sheet. It also captures what arrives from outside (an email from the supplier warning of a viscosity change) at second zero.
• Agent — cross-references the web reading with calender temperature, the polymer batch and the history of known drifts. It detects the drift before the reading crosses the SPC limit, and proposes the adjustment (correct the extruder screw, drop the calender by 2°C, adjust suction). It does not act alone — it proposes, the person validates and signs.

See the full IRIS architecture →

Before and after

Lab sampling vs. in-line control with iLEAN Edge

Aspect	Lab sampling + classic SPC	With iLEAN Edge + AI agent
Measurement frequency	Every shift or every finished roll	Continuous, over the web
Drift detection	Once the limit has already been crossed	Anticipated by cross-referencing signals
Root cause	Investigated after the fact	Polymer batch + temperature + speed cross-referenced
Rolls out of tolerance	Detected at weighing / at the customer	Held before palletizing
Operation without a network	n/a	Edge keeps measuring and alerting
Per-roll dossier	Rebuilt by hand from MES + lab	Automatic dossier with curve, batch and signature

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.

• Spunbond line for hygiene, medical or filtration, 1.6-3.2 m width, target basis weight in the 10-100 g/m² range.
• Edge pilot on 1 line (integrated web sensors + cross-referencing agent + integration with MES/production reporting). First value expected within a few weeks: the first early alert that saves a roll from running out of tolerance.
• Indicative payback between 4 and 9 months depending on line width and speed, the value per kg of nonwoven, the documented frequency of rolls rejected or reworked for basis weight and the average cost of a claim.
• Reduction in out-of-tolerance rolls on the order of ≥30% already during the pilot, without touching the physical set-up of the line.
• The hard lever is threefold: rejected rolls avoided, fine-tuning of the set-point (you run closer to the minimum basis weight and save polymer) and roll-by-roll traceability for the customer.

And the production manager's reasonable doubt

“What if the agent makes us change parameters for nothing?” — the agent proposes, it does not act. The iLEAN piece never touches the calender or the extruder directly: it puts the adjustment on the table of the operator or the line lead, with the curve and the probable cause. And reliability when the task is anchored to the source (sensors + parameters + batch) brings error below 1.5% ^[1]. The three safety rings exist precisely for this: ring 1 (critical OT) accepts nothing the person has not signed.

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