The losses that never make it into the monthly review
Packaging plants measure waste. What they rarely have is waste with a cause attached to it, on the shift it happened.
Waste is a total, arriving at month end
You know the percentage. You cannot say how much of it was startup, how much was running, how much was trim, and which jobs and substrates produced it.
Makeready varies two-fold between shifts
The same job, the same press, twice the setup time depending on who is on. Without press signals marking makeready start and end, that variation stays invisible and unmanaged.
Lines rarely run at rated speed
Chronic speed loss is the most under-measured loss in the industry. Ten percent below rated, every shift, on every job, compounds into a line you never needed to buy.
Web breaks stop the line without warning
The tension and drive-current patterns that precede a break are recorded by the drives and read by nobody.
WIP disappears between stages
Extrusion, printing, lamination, slitting and pouching each have their own logs. The flow between them, and where the material is right now, lives in someone's head.
A complaint can't be traced back to a reel
When a customer returns material, tying it back to the reel, the shift and the process conditions that made it takes days — if it is possible at all.
What we connect to in your plant
Multi-stage plants are the hard case, and the reason a generic IoT platform struggles here. We model the flow between stages, not just the machines.
Controls and drives
Business systems
The signals we read, and what each one tells you
The signals matter less individually than in combination — speed against substrate, tension against break, waste against the job that produced it.
| Signal | What it tells you |
|---|---|
| Line speed against rated speed, by job and substrate | The real speed-loss picture, separated from downtime, which is where chronic capacity quietly disappears. |
| Metres produced and metres wasted, split by startup, running and trim | Waste with a cause instead of a monthly percentage. |
| Web tension, unwind and rewind diameter, drive current | The pattern that precedes a break, and the material handling behaviour behind it. |
| Registration error and colour deviation | Print quality drifting while the job is running, rather than at the QC table afterwards. |
| Makeready start and end taken from press signals | Setup time nobody has to log by hand, and therefore setup time nobody can round down. |
| Stop events with reason codes | Downtime split cleanly across mechanical, job change, material and quality. |
| Reel ID joined to process conditions | Traceability from a customer complaint back to the exact run, shift and settings. |
| Dryer zone temperatures, ink, solvent and adhesive consumption | Recipe adherence and consumable cost against BOM, per job. |
| Energy per thousand square metres | Conversion cost by line and by SKU, which most quoting still treats as a flat overhead. |
One picture from extrusion through to dispatch
Stage-level numbers that roll up to a plant view, so a slitting problem stops being blamed on printing.
OEE by line and by stage
%, live
Waste split by cause
% startup / running / trim
Makeready time per job
minutes, by press
Speed loss against rated
%, by substrate
Metres per hour
by line and job
Roll yield
% good metres per reel
Scrap value per shift
currency
Changeovers per shift
count, by line
Energy per 1,000 m²
kWh
Ink and adhesive vs BOM
% variance
WIP between stages
tonnes or reels
OTIF by customer
%, monthly
AI trained on your process. Not a generic factory.
Models built around web processes and short-run, high-SKU scheduling — the two things that make packaging different from discrete manufacturing.
Waste attribution
Splits every metre of waste across cause, job, substrate, machine, operator and shift, so the three biggest drivers appear on one screen instead of being argued about in the production meeting.
Makeready prediction and job sequencing
Learns what each changeover actually costs on each press, then orders jobs by substrate, width and ink set to cut total setup time across the shift rather than optimising one job at a time.
Web-break early warning
Watches tension, roll diameter and drive-current patterns for the signature that precedes a break, and alerts the operator with enough lead time to slow down instead of losing the web.
Speed setpoint guidance
Recommends the fastest speed that has historically held quality for that substrate, job and machine — replacing the speed the previous shift happened to leave on the panel.
Your first thirty days
One line, end to end, is more useful than one machine on five lines. We start where the flow is.
Connect one line end to end
A single printing or extrusion line including its downstream slitting, tapped at the drives and controls. Older presses get retrofit counters and encoders.
2
Baseline waste and makeready
Waste split into startup, running and trim, with makeready timed automatically. This is usually the point at which the first argument gets settled by data.
3
Extend across stages and link reels
Adjacent stages come on, reel IDs are carried through, and traceability from complaint to run becomes a query rather than an investigation.
4
Show the money on one product family
One SKU family, costed properly — waste, speed loss, setup, energy — in currency. That is the number that justifies the rollout.
What plants aim for in the first quarter
15–30%
waste reduction on the connected linesp;
20–40%
makeready time reduction from sequencing
5–12%
throughput gain from recovering chronic speed loss
minutes
to trace a complaint back to a reel, instead of days
Target ranges for scoping a pilot. Replace with your own validated customer results before this page goes live.
