A digital factory for high-mix low-volume production

Many product variants, small batches and frequent changeovers: high-mix low-volume production takes more than a flexible machine. Operators need to know which instructions apply, planners need up-to-date progress information and quality teams need to trace what has happened to a product. When that information is scattered across paper, spreadsheets and standalone systems, every changeover means more searching and checking.

A digital factory connects machines, software and production data. Not to automate every decision, but to give people the same, up-to-date information to work with. At Vlegel Technology in Genk, we approach this as a software engineering team for machines, data and AI: what information is missing on the shop floor, and which connection would make it useful?

Start with the order and the workstation

With small batches, machine status is only part of the picture. You also need to know which order the machine is running, which product variant is being made and which operation comes next. Without that context, a counter reading or downtime signal tells you little about progress.

So connect order data from the ERP system to work on the shop floor. Digital work instructions need to match the correct variant and revision. Also record which materials, operations and checks belong to each product. That makes traceability part of the workflow, rather than a reconstruction after the event.

Connecting machines with OPC UA and MQTT

Machine data can, for example, be read via OPC UA and made available to other applications via MQTT. The right approach depends on the machine controller, available interfaces and existing network. An older machine may need a different connection from a new production line.

A Unified Namespace (UNS) can provide a shared, logically structured data model. Machines and systems publish their data under agreed names, so applications use the same meaning for a machine, order or status, for example. The UNS does not replace the ERP system. It helps make data usable across systems.

Those conventions are at least as important as the protocol. Define statuses, units, timestamps and data sources. Make it clear when a connection drops, too: the last value received does not necessarily reflect the current state.

Measure OEE, but keep it in context

OEE highlights losses in availability, performance and quality. In high-mix low-volume production, that figure needs careful interpretation. Changeover times, varying cycle times and small quantities affect comparisons between orders. One overall figure does not tell you why a process is slowing down.

So look at the underlying events too: when did the changeover start, when was the first good product made and why did the machine stop? Combine automatic logging with targeted input from operators. Use the data to investigate causes, not to judge performance without context.

Digital instructions in practice

At e-power, the challenge was starting up a new production line with many variants and small batches. Paper instructions made it difficult to work efficiently and without errors. Digital work instructions, linked to the Odoo ERP system, provided real-time insight into performance and reduced the risk of errors. New staff could also be trained faster.

The key lesson is the link between instructions and production context. A digital version of a paper document is not enough on its own. People need the right information at the right time.

AI reads along, a human decides

With a reliable data foundation, AI can help summarise production information or flag anomalies for investigation. We keep the boundary clear: AI only reads and makes suggestions; a human decides. AI does not independently change machine parameters, orders or quality approvals.

Start with one clearly defined problem, such as missing downtime reasons or incorrect instruction versions. First check that the data is accurate and the solution works on the shop floor. Want to discuss where to start with your factory or machine? Book a 15-minute call with Vlegel; we will look at the information and connections you need.