2K dosing of MS polymer sealant — a badly measured ratio is a pallet of cartridges to destroy.
The base-to-catalyst ratio on a 2K MS polymer dosing unit shifts because of bubbles in the catalyst, thermal drift in the drum or worn seals — and the machine builder's flowmeter never sees it. iLEAN Edge adds vision on the bead, an auxiliary thermal probe and the pump's vibration pattern, stops the line on drift and alerts the manager. The person signs — the dosing unit does not restart on its own.
The flowmeter says everything is fine — and the cartridges cure badly.
A 2K MS polymer dosing unit runs two pumps in parallel (base and catalyst) and a static mixer at the end. On paper, the nominal ratio is settled: so many grams of A for so many grams of B. In the plant, the effective ratio moves through routes the PLC never records:
- A bubble in the catalyst — the drum empties, the suction line pulls air, the flowmeter keeps counting volume but it is no longer pure catalyst.
- Thermal drift in the base drum — viscosity rises at night, falls during the day, pumping changes and the static mixer no longer mixes the same way.
- A check valve with play — catalyst creeps into the base line between shots and skews the next cartridge.
- A catalyst batch change — it arrives with properties inside spec but different; cure time moves without anybody touching a single setpoint.
The operator finds out when the first cartridge shows a tack-free time that is too long, or when a customer returns a batch for out-of-range green strength. By then dozens of bad cartridges have already been filled — raw material cost, destruction cost and, worst of all, brand cost if they reach the end user.
iLEAN Edge does not replace the PLC — it seals the crack between what it measures and what actually happens.
The problem with 2K mix control is not a lack of sensors on the dosing unit — it is that the sensors that come from the factory measure volume and pressure, while the truth about the ratio lives in the bead, in the exotherm and in the vibration signature of the pump. iLEAN acts as the putty that covers the dead zone between the flowmeter and the cured cartridge, without asking you to change the dosing unit or renegotiate with the machine builder.
Edge sees the bead at the mixer outlet. Connect reads the catalyst batch and the drum viscosity. The agent cross-references them with the recipe and the expected curve — and if something does not add up, it stops the dosing unit and opens the bypass. The person signs — never the other way round.
The three iLEAN pieces applied to 2K mix control on MS polymer:
- Edge — a terminal with machine vision (CNN) over the static mixer + an auxiliary thermal probe at the cartridge head + an accelerometer on the dosing pump. It reads the color and symmetry of the bead, the first exothermic curve and the pump's vibration signature. If the three readings do not match the learned pattern, it fires the actuator (stop + bypass) in milliseconds. It works with no network.
- Connect — captures the technical data sheet of the catalyst batch (the supplier's PDF by email), the drum viscosity measured with the handheld viscometer (a photo of the display from the operator's phone), the ambient temperature of the warehouse. At second zero, with nobody having to resend anything.
- Agent — cross-references the nominal recipe with the catalyst batch, the measured viscosity of the base drum, the exothermic curve of the first samples and the history of the dosing unit. If the drift has a probable cause, it proposes it to the manager with the concrete data point — it does not send an email at 10 pm, it opens the conversation through whichever channel that person uses.
Manual 2K control vs. cross-checked 2K control with iLEAN
| Aspect | Flowmeter + operator's eye | With iLEAN Edge + Connect + agent |
|---|---|---|
| Reading the ratio | Volume — does not detect entrained air | Bead + exotherm + pump vibration |
| Catalyst batch change | Data sheet on paper, filed away | Cross-referenced with the recipe and the expected curve |
| Thermal drift in the drum | Viscosity assumed constant | Real viscosity cross-referenced with the setpoint |
| Detecting an off-recipe cartridge | Later QC or a customer complaint | Stop and bypass before the next shot |
| Operation with no network | n/a | Edge keeps stopping the line on the panel's own power |
| File for an IATF auditor / OEM supplier audit | Rebuilt by hand | Per-shift dossier with a photo of the bead and the curve |
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.
- MS polymer sealants plant, 2 parallel 2K dosing units, multi-recipe with tin and titanium catalysts, filling 290 ml cartridges and 600 ml sausage packs.
- Edge pilot on one dosing unit (camera over the static mixer + auxiliary thermal probe + accelerometer + integration with the machine PLC). First value expected within a few weeks.
- Indicative payback between 4 and 9 months, depending on the frequency of documented out-of-range cure times and the average cost of destroying/reworking cartridges.
- Expected reduction in off-recipe cartridges of ≥ 30% in the first quarter, based on incidents avoided.
- The hard lever is a single run saved: cartridges recovered, raw material saved by bypassing in time, an OEM complaint avoided.
And the process manager's reasonable doubt
"What if the AI gets it wrong and stops the dosing unit for no reason?" — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI merely compares what it sees against a learned pattern (bead color vs. reference bead, exotherm vs. expected exotherm), the best models brought error below 1.5% [1]. And even so, what is critical is never decided alone: iLEAN stops and bypasses, the person signs the restart. The three safety rings exist precisely for this.
[1] OpenAI paper "Why Language Models Hallucinate", 2025 — on the reliability of AI in anchored tasks.
What people ask about 2K mix control on MS polymer
Why does the base-to-catalyst ratio drift on a 2K MS polymer dosing unit?
2K dosing units run two pumps in parallel — the MS polymer base and a tin or titanium catalyst — with a static mixer at the end. Ratio drift shows up through several routes: entrained air in the catalyst, a viscosity change caused by the ambient temperature of the drum, worn seals, a check valve with play, the end of a catalyst batch with different properties. The operator sees it once there are already cartridges curing to the wrong time — in cure time, in green strength, in tack-free time. Not before.
What does iLEAN Edge see on the dosing unit that the PLC does not?
The PLC sees pressure, flow and setpoint. iLEAN Edge adds three readings that no sensor supplied by the machine builder delivers: an image of the bead at the static mixer outlet (color, symmetry, presence of micro-bubbles), the exothermic curve of the first samples from every cartridge filled, taken with an auxiliary thermal probe, and the vibration pattern of the dosing pump. With those three, Edge concludes whether the effective ratio has moved — even if the flowmeter keeps reading exactly what it always reads.
What does the system do when it detects a ratio drift?
Edge fires two actions in milliseconds: it stops cartridge filling and opens the bypass to the purge container. In parallel, an iLEAN agent alerts the process manager through whichever channel they use (radio, phone with an earpiece, tablet), with the probable cause inferred — a bubble in the catalyst, thermal drift in the base drum, a fatigued check valve. The person validates and signs; the dosing unit does not restart on its own. Edge does not decide what to change, it only prevents more bad cartridges from coming out.
Does Edge work if the plant loses its network?
Yes. Detection and the stopping of the dosing unit live inside Edge itself — the CNN is embedded, the actuator is a dry contact. As long as the panel has power, the see-and-stop cycle keeps running. What needs the network is traceability upstream (sending to the MES, the dossier to the agent), and that accumulates locally and uploads when the link comes back. It is an axiom of serious OT architectures: what is critical cannot depend on WiFi.
What is the return on a 2K mix control pilot in specialty chemicals?
The order of magnitude of an Edge pilot on a 2K dosing unit is in the same range as any Edge inline inspection pilot: an initial investment covering the camera, the auxiliary thermal probe, the actuator and the integration with the PLC, plus an annual license. The hard levers are off-recipe cartridges avoided (rework or destruction), raw material saved through the bypass when drift is caught in time, and customer complaints over out-of-range cure time. We ask for your data and send you the estimated ROI in 48h.
Tell us about your case and in 48h we'll send you the estimated ROI of this AI project for your 2K dosing unit.
We work on your plant's real data, not ours. Diagnostic with no commitment.
Request estimated ROI in 48h See specialty chemicals