Detergent formulation with AI — the batch that separates in the warehouse never showed its face in the reactor.

The problem

The batch the lab approves is not always the batch that survives six weeks on the shelf.

Liquid or powder detergent formulation lives in the balance between what development says should work and what the plant and the supply chain actually do with that formula. Three realities, three different owners:

1. The real variability of raw material — the batch of LAS, ethoxylates, enzymes or nonionic surfactants is not always the one used in development. The supplier certificate arrives by email or as a PDF, and the veteran formulator adjusts by eye.
2. The real reactor profile — addition order, agitation energy, jacket temperature, enzyme addition window. It lives in the reactor's SCADA when there is one, and on an operating sheet when there is not.
3. Packing and storage conditions — ambient temperature, time from batch to filling, humidity. Whether a batch separates at six weeks usually comes down to something that happened (or did not happen) in those three hours.

The formulation manager knows this, but cannot be in all three places at once. The reactor operator has the recipe and the SCADA screen. And the batches that separate on the shelf show up as a customer complaint, not as a plant event. The classic system works 96% of the time. That 4% is the batches that get reformulated, reworked or returned — and each one pays for the quarterly committee meeting.

How it fits the IRIS system

iLEAN does not reformulate your detergent — it seals the cracks between lab, reactor and packing.

The problem with detergent stability is not a lack of instrumentation: it is information living in islands (batch certificate, frozen recipe, real reactor profile, packing conditions, separation history) that never arrives cross-referenced at the critical moment. iLEAN acts as the putty that fills the gaps, without asking you to change your ERP, MES, reactors or lab.

Edge watches the dissolution and the packing. Connect captures the batch certificate and the recipe wherever they live. The agent cross-references the three realities and proposes the adjustment before packing. The person signs — no batch releases itself.

The three iLEAN pieces applied to detergent formulation:

• Edge — a terminal with vision (CNN) and analog inputs on the reactor and the packing line. It detects abnormal foam, early separation, out-of-band turbidity, continuous viscosity, provisional color and appearance drift. It reads the real jacket temperature, the pressure gauges and the dosing scales. If the pattern is clear, it holds the batch before packing. It works without a network.
• Connect — captures the certificate of every surfactant and enzyme whether it arrives by EDI or as a PDF in the technical manager's inbox, the batch recipe in the MES or the formulator's Excel, and the supplier's historical variability. Whatever arrives from outside (a retailer changing its standard, new regulation, a new enzyme batch with a different declared activity) enters at second zero.
• Agent — cross-references the batch certificate, the recipe, the real reactor profile and the Edge reading. It recognizes the pattern ("with this LAS batch and this jacket temperature, the enzyme SKU tends to separate at 6 weeks if warehouse ambient temperature exceeds X°C") and proposes the adjustment before packing. It builds the per-batch dossier for the ISO 9001/14001 auditor and for the retailer. The person validates and signs.

See the full IRIS architecture →

Before and after

Formulation steered by complaints vs. cross-referenced with iLEAN

Aspect	SCADA + lab + formulator's eye	With iLEAN Edge + Connect + Agent
Surfactant variability	Certificate by email, stays in purchasing	Cross-referenced with the formula and the reactor profile
Enzyme addition	Recipe + clock + operator's eye	Real window verified against the jacket curve
Separation detection	Customer complaint at 6 weeks	Anticipated from historical patterns before packing
Veteran formulator's knowledge	Lives in their head	Captured as a persistent agent pattern
Operation without a network	n/a	Edge keeps reading and holding
Per-batch dossier	Rebuilt by hand when a complaint arrives	Automatic traceability with appearance photo and certificate

Impact estimate

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

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

• Liquid or powder detergent plant, multi-SKU, multi-customer (consumer + professional + retail), with recurring complaints for separation or out-of-spec viscosity.
• Edge pilot on one reactor + packing, Connect for the batch certificate and the recipe, agent cross-referencing them. First value expected within a few weeks.
• Indicative payback between 4 and 9 months, depending on the share of reformulated/reworked batches, the average cost of rework and the volume of customer complaints.
• The hard lever is threefold: reformulation reduced ≥30%, fewer returned lots and better use of surfactants and enzymes.

Why this is the moment for a chemical plant

Between 2000 and 2021, the least digitized sectors barely improved productivity; the most digitized raised it by up to 40% ^[1]. Consumer chemicals is one of the sectors where that gap is opening fastest — because unit margin is tight and the difference between operating blind and cross-referencing sources shows in every batch. iLEAN closes that gap without asking you to change the plant: it leverages what you already have.

^[1] Source: BBVA Foundation / Ivie — productivity by sector digitization level, 2000–2021.