Connect any machine: OT/IT integration without a heavy project

The false obstacle of old machines

“Our machines are too old to be connected.” You hear that sentence in nearly every plant, and it blocks countless performance initiatives before they even begin. The underlying idea is that a machine with no digital interface is beyond the reach of any measurement. It is a widespread belief, and a largely unfounded one.

Because connecting a machine to measure its OEE does not mean integrating it into a complex information system. A sensor fitted directly on the machine captures the essentials: stops, speed and, where relevant, quality. It doesn’t need the machine to be communication-capable, nor to have a recent PLC. The measurement happens as close as possible to the physical reality, which makes old machines just as reachable as new ones.

OT and IT: two worlds you shouldn’t pit against each other

Connecting machines brings two universes into play: OT, the operational technology of the shop floor (PLCs, sensors, equipment), and IT, the management information systems. Big integration projects try to marry the two deeply, which is complex, costly and often a source of friction between teams with different cultures.

The standalone measurement-layer approach sidesteps that difficulty. It does not try to fuse OT and IT, but to capture performance data where it is born, on the machine, without imposing a rework of the industrial network or deep integration. The OT/IT friction that sinks so many projects is thus avoided: you measure first, without waiting for the two worlds to be perfectly aligned. The deep integration, if it is ever needed, can come later and on its own schedule, once the measurement has already proven where the losses are and what they cost. You get the value now and keep the bigger architectural decision for when you actually have the data to justify it.

With or without a standard protocol

When a modern machine exposes its data through a standard protocol such as OPC UA, you can lean on it to retrieve the information: that’s an efficient route when it is available. But it is not indispensable, and that is the essential point. Many industrial fleets mix equipment of very different ages and brands, only some of which speaks a standard protocol.

Tying measurement to the presence of OPC UA would mean excluding a large part of the fleet, precisely the oldest machines that often harbour the most losses. With a standalone sensor, that constraint disappears: with or without a standard protocol, you can measure. The solution adapts to the fleet as it is, not the other way round.

What the sensor actually captures

For OEE, the essentials come down to three pieces of information: is the machine running or stopped, at what speed is it producing, and is the output conforming. A well-placed sensor picks up these signals directly, continuously and to the second. That is enough to reconstruct a complete OEE measurement, including the losses a manual log never sees.

This economy of means is a strength. The aim is not to instrument everything or pull back every machine variable: you capture what matters for performance, reliably and continuously. It is that focus that allows a light, fast installation while producing rich, usable data for improvement. From those three signals you reconstruct the full OEE breakdown, Availability times Performance times Quality, with each loss attributed rather than lumped into a single vague figure, which is exactly what a manual reading can never deliver.

From collection to value

Connecting a machine and collecting its data is only one step. Value does not arise from the collection itself, but from what you do with it. Data that is recorded but never looked at improves nothing. That is why the rendering matters as much as the capture: performance data has to be visualised in real time, where the teams work.

Displayed on the floor, fresh and legible, it becomes an immediate decision tool. The team sees the stop the moment it happens, understands the pace drift, acts, and checks. Without that loop, connecting machines would be a purely technical exercise. With it, the connection becomes a concrete lever for steering performance.

Security and a controlled scope

A legitimate worry surrounds any machine connection: opening a security hole or making the system more complex. That is exactly where a standalone layer reassures. By touching neither the machine’s PLC nor the MES, it keeps the integration surface, and therefore the associated risks, to a minimum. The scope stays controlled and contained.

That lightness is also a guarantee of reversibility. You can start on one machine, observe, and extend without having committed to any irreversible transformation of the information system. The connection is made by adding a measurement layer, not by overhauling the industrial core, which limits both technical risk and the coordination effort between teams. Because the layer sits outside the control loop, it cannot disturb the machine’s operation: it reads, it does not command, and that read-only posture is what makes IT and automation teams comfortable signing off on a rollout that would otherwise demand months of security review.

A heterogeneous fleet is not an obstacle

Most plants live with a heterogeneous fleet: machines of different brands, ages and generations, some recent and communication-capable, others old and silent. That diversity often discourages classic integration projects, which assume a degree of homogeneity or demand specific adapters for each type of equipment. Heterogeneity then becomes a brake, even a pretext for inaction.

With a standalone measurement layer, that diversity stops being a problem. The sensor adapts to the fleet as it is, machine by machine, without imposing a common standard. You can instrument an old press and a recent line the same way, and obtain a homogeneous, comparable measurement from equipment that is nonetheless disparate. It is that ability to work with the existing fleet that makes the approach realistic in the real life of shop floors, where the ideal fleet never exists.

Start small, extend with confidence

The right connection strategy is gradual: one line first, to prove the value, then extension to the rest of the fleet once the approach is proven. Each machine connects with the same lightness, which makes the rollout linear and predictable, without the tunnel effect of big integration projects.

That gradual pace protects the plant from risk and leaves it in control of the tempo. You move forward according to priorities, building at each step on the experience gained. More than 450 plants across 30+ countries already monitor their OEE to the second with TeepTrak. Connecting a whole fleet then becomes a series of simple installations rather than a monolithic undertaking, and performance gains visibility as the rollout extends.

Key takeaways

Connecting your machines to track performance does not require a big OT/IT integration project. A standalone sensor fitted on the machine captures stops, speed and quality, on old and new equipment alike, with or without a standard protocol. The data, visualised in real time on the floor, becomes a decision tool. The scope stays controlled, and you start small before extending, with usable data within 48 hours. A heterogeneous fleet, far from being an obstacle, is measured this way in a homogeneous and comparable manner, machine by machine, without imposing a common standard or waiting for OT and IT to be perfectly aligned.

FAQ

How do I connect old machines to performance tracking?
With a standalone sensor fitted directly on the machine, capturing stops, speed and quality with no PLC, SCADA or digital interface required. The measurement happens as close as possible to the physical reality, which makes old machines just as reachable as new ones.

Do I need OPC UA to connect my machines?
No. OPC UA is useful when it exists, but not indispensable: a standalone sensor measures with or without a standard protocol. Tying measurement to OPC UA would exclude a large part of the fleet, precisely the oldest machines.

Do I have to integrate OT and IT deeply?
No. The measurement layer is standalone: it captures data where it is born, on the machine, with no industrial-network overhaul or deep integration. The OT/IT friction that makes so many projects fail is thus avoided.

Does the connection pose a risk to the system?
The risk is minimal: by touching neither the PLC nor the MES, the measurement layer reduces the integration surface. The scope stays controlled, reversible, and you can start on one machine before extending.

How long does it take to capture the data?
The first usable data arrives within 48 hours, on any kind of fleet, after a sensor install of under an hour.