Every OEE program eventually runs into the same wall: the machines. Some are new and chatty, some are decades old and silent, and a few have no PLC at all. This guide gives OT and IT a single integration playbook that captures clean signals from any asset and routes them to overall equipment effectiveness in days, not months.
Why integration, not instrumentation, is the real project
Most teams frame an OEE rollout as a sensor project. The harder, and more valuable, work is machine data integration OEE: turning a mix of PLC tags, discrete sensors and silent legacy assets into one consistent stream of states, counts and reasons. A line is only as measurable as its least-connected machine, so the integration strategy has to cover the whole asset mix, not just the easy new cells.
Manual logs are the default fallback when integration looks hard, and they quietly corrupt the number. Hand-written stop sheets typically overstate OEE by 8 to 15 points because operators round, batch and forget. Automatic capture removes that bias at the source, which is why the integration design matters more than any single dashboard feature.
The three signals every machine must yield
- Run state: is the asset producing, stopped, starved or blocked, captured continuously.
- Good count: cycles or units produced, separated from scrap and rework.
- Reason: a structured stop reason, so downtime maps to the Six Big Losses, not a free-text guess.
Three capture methods for one mixed floor
There is no single right way to connect a machine. Modern assets expose tags over an industrial protocol and can be read directly. Older assets often expose nothing useful, so a discrete sensor on the cycle, the stack light or the output gives a clean, vendor-neutral signal. And where there is no PLC at all, an edge device such as the TeepTrak Box reads sensors locally and computes states without touching the machine controller.
The art is matching method to asset so that data trust stays high and install time stays low. The comparison below is the working decision table OT and IT can apply machine by machine.
| Capture method | Best fit | What it reads | Install effort |
|---|---|---|---|
| PLC tag read | Modern assets with open protocols | States, counts, alarms from controller tags | Low to medium |
| Discrete sensor | Legacy or closed controllers | Cycle, output or stack-light signal | Low |
| No-PLC edge (TeepTrak Box) | Assets with no PLC or no access | Sensor inputs computed at the edge | Low, no controller change |
Hutchinson, a Tier-1 automotive supplier, raised OEE from 42 to 75 percent, a gain of 33 points, while standardizing capture across 40 sites in 12 countries.
Connect one line and watch the data arrive
Run a free 60-day pilot on one line. We connect the mixed asset base and surface the first losses within about two weeks.
Download the OT/IT integration guide
The capture-method decision table, the edge architecture and the security checklist. We send it to your work email.
Edge architecture and security by design
A sound architecture keeps capture local and the data path clean. An edge device computes machine states on the floor and forwards only structured events, which keeps bandwidth low and avoids exposing controllers. The TeepTrak Box sits at the edge with no PLC dependency, while PerfTrak turns those events into real-time OEE that operations and IT see together.
Security follows the same principle. Capture is read-only where possible, network segments stay separated, and the integration adds no inbound path to the control layer. IT keeps governance over identity, transport and storage, while OT keeps the machines untouched. That split is what lets a rollout move quickly without raising the risk profile of the plant network.
From first connection to OEE in days
Heavy MES integrations measure in months because they try to model the whole plant before producing a number. An edge-first approach inverts that: connect one line, validate the three signals, and read trustworthy OEE almost immediately. The hidden factory, the 30 to 45 percent of capacity lost to unmeasured stops, micro-stops and speed loss, becomes visible as soon as the data is honest.
Once one line is proven, the same capture patterns replicate across the floor. Legacy machines reuse the sensor template, modern machines reuse the tag map, and no-PLC assets reuse the edge config. Integration stops being a bespoke project and becomes a repeatable rollout, which is how a program scales from a pilot to a plant without a multi-month freeze.
- Treat the rollout as integration across the whole asset mix, not instrumentation of the easy machines.
- Match capture method to each asset: PLC tag, discrete sensor or no-PLC edge.
- Keep capture local and read-only so security and uptime stay intact.
- Connect one line first and read trustworthy OEE in days, then replicate the patterns.
0 Comments