Where Factory Data Comes From: PLC, OPC UA, MES, ERP

Factories are full of data. The problem is rarely absence — it’s that machine status sits on one screen, production figures on another and alarms on a third. Some of it still lives in Excel files.

The first step of any digital twin rollout is mapping this. Below are the typical sources and what each one actually tells you.

PLC and SCADA — the pulse of the floor

A PLC (Programmable Logic Controller) is the controller that runs the machine. The rawest, most real-time data lives here: is the machine running, which step is it on, what’s the cycle counter, which alarm is active.

SCADA is the layer above that collects those signals and shows them to operators. It’s already installed in most plants.

Most of the core signals a digital twin needs come from this layer: state, counters, downtime reasons. The “do we need new hardware?” discussion usually ends here — the data is already being produced; it just needs to be brought out.

OPC UA — the common language

Machines from different vendors speak different languages. OPC UA (Open Platform Communications Unified Architecture) exists to solve exactly that: platform-independent, with a security layer, and widely adopted across the industry.

In practice it means this: even if your machines come from different manufacturers, if they support OPC UA they can be read through one interface. Asking suppliers for OPC UA support on new installations saves a great deal of work later.

MQTT — lightweight and publish-based

MQTT is a lightweight publish-subscribe protocol that works well on limited bandwidth. It’s particularly efficient for collecting data from many endpoints and for moving it between the floor and the cloud.

It isn’t an alternative to OPC UA but a complement: OPC UA defines the structure of machine data, MQTT is an efficient way to transport it.

MES — the layer that runs the floor

MES (Manufacturing Execution System) manages work orders: which product, on which line, in what quantity; traceability records; operator assignments.

For a digital twin, the value of MES is context: the PLC tells you the machine is running, MES tells you which work order it was running at the time. Put together, questions like “which product are we losing most on?” become answerable.

ERP — the plan of the business

ERP (Enterprise Resource Planning) manages orders, inventory, purchasing and finance. It sits far from the floor, but you need it to compare planned against actual.

MES and ERP are often confused. In short: MES answers “what is being produced now”, ERP answers “what was planned to be produced”.

SQL databases and Excel

In many plants, critical information lives in a SQL database or an Excel file in a shared folder: shift patterns, changeover times, target quantities, cost rates.

Dismissing these is a common mistake. Leaving Excel outside the system often means leaving out the most important input to the calculation. A digital twin can read those files too.

Do you need all of it?

No. To start, the PLC/SCADA layer is usually enough: with state, counters and downtime reasons you can compute OEE and produce a live view of the line. As MES, ERP and cost data are added, the model deepens — losses can be converted from units into money, and planned can be compared against actual.

A practical note: don’t try to know exactly what data you have before you begin. The discovery phase produces that map anyway, and it usually finds more than expected.

---

OnoTwin Studio is built to connect to all of these sources; our digital twin solutions page walks through how a rollout runs from discovery to commissioning.