Skip to Content
TutorialsPreventive Maintenance

Preventive Maintenance

Unplanned downtime is the most expensive kind. Most failures announce themselves: a bearing gets warm, a motor starts vibrating, a pump loses pressure. If you measure those signals, maintenance can be planned instead of rushed.

This tutorial sets up condition monitoring for a machine: sensors, trends and alerts.

What to measure

Machines rarely expose their condition by themselves. Two ways to get the data:

  • From the PLC — many machines already measure temperatures, pressures and motor currents. Read them via OPC UA or Modbus.
  • Add sensors — an IFM IO-Link Master makes it easy to add vibration, temperature or pressure sensors to an existing machine, without touching the machine’s own control.

Start small: one temperature or vibration sensor on the machine that hurts most when it fails.

Collect the values

Create the connection and a rule per value, for example motor_vibration and motor_temp. On the Deduplication transformation, set a Bandwidth so small sensor noise is not stored, and keep the Timeout so you can see the sensor is alive. Add a Filter transformation to drop impossible values from sensor glitches.

Track operating hours

Wear follows running time, not calendar time. If the machine has an on/off signal (or you create one with the Boolean transformation), add a Timer transformation to measure runtime. In a dashboard, the running hours per week tell you when the next service is due.

Build the trend dashboard

Create a dashboard with a Time series panel per condition value, aggregation Time-weighted average, bucket per day, over a range of months. Condition problems show as slow trends: this week’s value is normal, the three-month trend is not.

Alert before it fails

Create alert rules:

  • Threshold — vibration or temperature above the level that is normal for this machine. Look at a few weeks of data first to know what normal is.
  • No data — fires when the sensor stops reporting. A dead sensor is also a maintenance job, and without this alert nobody notices.

Send the alerts to an output your maintenance team actually reads, for example a Telegram group or a team email address.

💡

Maintenance also becomes measurable the other way: track downtime as operations with the machine’s on/off signal, and see whether your maintenance work actually reduces the number and length of stops.

Last updated on