Initial Situation

The client, a specialized machine manufacturer with 120 employees in southern Germany, produced custom automation solutions. Production ran across three shifts, with machines frequently failing in the early morning hours. The IT department only reacted once production had already stopped and the shift supervisor had been manually informed.

  • Reactive Approach: Failures were only detected after they had occurred.
  • Data Silos: Machine data was stored in local SCADA systems and could not be analyzed centrally.
  • High Downtime Costs: Each unplanned hour of downtime cost the operation an estimated €1,500 in lost capacity and delay risks.

Implementation

We implemented a central monitoring system that securely transfers OT data into the IT environment and visualizes it. The architecture was based on a robust open-source stack that required no high licensing costs and was hosted entirely in the cloud.

  • Data Acquisition: Installation of edge gateways at critical CNC and robot cells to capture status, temperature, and vibration data.
  • Metric Storage: Deployment of Prometheus as the central time-series database for all machine telemetry data.
  • Visualization: Creation of Grafana dashboards that consolidated real-time status, OEE metrics, and historical trends into a single interface.
  • Alerting: Configuration of intelligent threshold alerts that warn IT and the shift supervisor 15 minutes before a predicted failure.

The implementation was completed within six weeks in three phases: piloting on one line, rolling out to all production areas, and training the specialized departments.

Result

Within the first three months after the rollout, a significant decrease in unplanned downtime was observed. The team was able to perform preventive maintenance during the warning periods before critical components failed.

  • Downtime: Reduction of unplanned failure times by 40% compared to the same period in the previous year.
  • Response Time: Shortening of the time between fault detection and initial response from 45 minutes to under 10 minutes.
  • OEE Increase: Rise in Overall Equipment Effectiveness (OEE) from 72% to 81%.
  • Transparency: For the first time, there was a unified data basis for production, maintenance, and IT.

The investment in the monitoring system paid for itself within just four months through avoided downtime costs and higher throughput rates.

Learnings

The success lay not only in the technology but in the close collaboration between IT and production. The dashboards had to be easily readable for shift supervisors, not just for engineers.

  • User-Centric Design: Grafana dashboards were specifically tailored to the roles of shift leadership and maintenance.
  • Training is Mandatory: Regular workshops significantly increased the acceptance of the new tools among the workforce.
  • Scalability: The Prometheus architecture can be easily extended to additional sites or new machine generations.

Results

  • -40% – Unplanned downtime
  • +9% – OEE (Overall Equipment Effectiveness)
  • < 10 Min – Average response time during failures