Getting machine data out of the system that collects it

A factory is rarely homogeneous. It stacks machines from different generations, each with its own format, protocol and data model. As long as those blocks exchange nothing, performance stays whatever each machine achieves on its own.

The test that costs five minutes

Ask your supplier, or the one you are evaluating, for an export of last month's readings in a format you can open without them. Three answers are possible: it is immediate, it is billable work, it is not possible. The third one tells you your machine data is not really yours.

Why it is rarely immediate

• Silos: every line, every system (ERP, MES, SCADA) runs for itself, and information stays captive where it was produced.

• Proprietary protocols: one manufacturer's machine does not natively speak to another's, or to an energy meter. Industrial history produced as many dialects as product ranges.

• The cost of translation: linking disparate systems takes bespoke development, slow to write and unpleasant to maintain.

Open standards exist and cut that cost. MQTT is light and suits carrying readings from sensors where bandwidth is limited and the link uncertain. OPC UA carries both the data and its description, from sensor through to processing. REST APIs, inherited from the IT world, act as the meeting point between the shop floor and management applications.

Transmitting is not understanding

Two temperature sensors can return the same number while measuring different things: different units, different ranges, different sampling rates. Without that context the value is ambiguous and any comparison is wrong.

This is what semantic interoperability addresses, and what OPC UA information models and digital twin work aim to establish. What it makes possible comes down to three things: a dashboard that aggregates several sources without distorting them, automation that compares thresholds which are genuinely comparable, and large-scale analysis that does not rest on implicit assumptions.

Four moves that open up a fleet

• Choose equipment compatible with open standards at purchase or at modernisation time.

• Isolate the translation layer in a gateway rather than scattering it across every application.

• Document and model your data flows so they stay manageable and auditable over time.

• Have IT and operations decide together; an architecture arbitrated by one side alone is paid for later.

Sycon works on heterogeneous fleets without modifying equipment. Access to readings goes through an open, documented and stable API that third-party systems (ERP, MES, maintenance management, business intelligence) query without a proprietary interface in the way. The data stays exportable and reusable by the customer, and the list of supported systems grows with the fleets encountered rather than being frozen in a catalogue.