What is Condition Monitoring?
Condition monitoring is the continuous or periodic measurement and analysis of parameters such as vibration, temperature, oil quality and current draw to determine the technical condition of machinery and installations and to detect developing faults before they cause a breakdown. It is a specific form of monitoring that focuses on the health of assets rather than on the process or the network. Condition monitoring supplies the measurements on which condition-based maintenance and predictive maintenance are built. The international framework is ISO 17359, the general guideline for setting up a condition monitoring programme.
By applying condition monitoring, organisations can reduce unplanned downtime, lower maintenance costs and extend equipment lifespan.
🧠 How does condition monitoring work?
A condition monitoring system turns a physical signal into a maintenance decision. The ISO 13374 standard divides that chain into six functional blocks. The MIMOSA specification OSA-CBM is an open implementation of the same model, so that systems from different vendors can exchange data.
| Functional block (ISO 13374) | What happens? | Where in the OT architecture? |
|---|---|---|
| Data acquisition | A sensor measures vibration, temperature, pressure, current or sound | Field level, for example via IO-Link, HART, WirelessHART or Bluetooth Low Energy |
| Data manipulation | Filtering, averaging, frequency analysis (FFT), calculating indicators | Smart sensor or edge device |
| State detection | Comparison with the baseline and alarm limits | Edge, SCADA or CM software |
| Health assessment | Diagnosis: which component, which failure mode, how severe? | CM software, historian, analyst |
| Prognostic assessment | Estimating how long the component will last | Analytics platform, machine learning |
| Advisory generation | Recommended action and timing | CMMS or MES for work order and scheduling |
Not every system runs through all six blocks. Simple temperature monitoring stops at state detection and only raises an alarm. Strictly speaking, prognosis belongs to predictive maintenance; the diagnostic techniques themselves are described in ISO 13379-1, a second edition of which was published in 2025.
🔄 What is the difference between condition monitoring, CBM and predictive maintenance?
Condition monitoring is often listed as a maintenance strategy, but it is primarily a measurement and analysis activity. The strategy that acts on it is condition-based maintenance (CBM): preventive maintenance based on an assessment of the physical condition.
| Type of maintenance | Trigger | Role of condition monitoring | Example |
|---|---|---|---|
| Corrective | The component has failed | None | Replacing a broken motor |
| Preventive | Fixed time or running hours | None or limited | Oil change every six months |
| Condition-based (CBM) | A measured value exceeds a limit | Provides the measurement and the alarm | Vibration rises above the alarm limit → bearing replaced |
| Predictive (PdM) | Forecast of the condition over time | Provides the trend for the forecast | Model predicts bearing damage within three weeks |
In short: without condition monitoring there is no CBM or predictive maintenance, but condition monitoring without follow-up in the maintenance process only produces charts.
🔧 Which forms and measurement methods are there?
Condition monitoring comes in three forms, which often exist side by side:
| Form | How it works | Typical for | Measurement interval |
|---|---|---|---|
| Periodic (route-based) | A technician walks a measurement route with a portable vibration meter, thermal camera or oil sample kit | Large numbers of pumps, fans and motors | Weekly to quarterly |
| Online (permanent) | Permanently mounted sensors with a continuous data stream to edge or server | Critical or hard-to-reach machines | Seconds to minutes |
| Machinery protection | Permanent monitoring that trips the machine on excessive vibration, for example to API 670 | Turbines, large compressors and pumps | Continuous, with trip function |
The main measurement techniques are vibration analysis, oil analysis, ultrasound, thermography and motor current analysis. Vibration analysis is the standard for rotating machinery; ISO 20816-3 gives zone limits in mm/s RMS. A detailed overview of techniques, standards and zones is given under predictive maintenance.
Many installations also monitor their own components. Smart field instruments report their health using the four status signals of NAMUR NE 107, and variable frequency drives supply motor current, torque and temperature. This “free” diagnostic information often goes unused because the DCS only reads the measured value.
🏭 Where is condition monitoring used?
- Rotating equipment - pumps, motors, fans, compressors, gearboxes and turbines
- Critical production equipment - conveyors, mixers, presses and packaging lines where a stop halts the whole line
- Process industry - valve diagnostics and field instrument monitoring, increasingly via the second channel of NAMUR Open Architecture
- Wind energy - most large modern wind turbines have a condition monitoring system for drivetrain and blades; DNV certifies such systems to its service specification DNV-SE-0439
- Buildings and infrastructure - pumping stations, locks, refrigeration plant and environmental conditions in cleanrooms
- Electrical installations - thermography of switchgear and transformer monitoring
Condition monitoring pays off most where downtime has a major impact on production, safety or the environment. The gain shows up in the availability factor of Overall Equipment Effectiveness and in a higher MTBF.
🧭 How do you set up a condition monitoring programme?
ISO 17359:2018 describes a fixed sequence. Translated into practice:
- Criticality analysis - rank installations by the consequences of failure, for example with an FMEA, and decide which assets justify monitoring. Align this with your asset management system to ISO 55000.
- Identify failure modes - for each critical machine: which defects occur and how do they become measurable? The failure mode determines the technique.
- Choose method and interval - select the technique, measurement point and interval. The interval must be well shorter than the time between a measurable incipient defect and failure (the P-F interval), otherwise the finding comes too late.
- Record a baseline - measure the healthy machine under known operating conditions such as load and speed.
- Set alarm criteria - combine standard values with the machine’s own trend. For a 15 to 300 kW machine on a rigid foundation, the boundary between zones B and C of ISO 20816-3 is 2.8 mm/s; ISO 20816-3 also treats a change of more than 25% of the upper limit of zone B as significant, even if the value is still below that limit. In practice, a doubling relative to the baseline is often taken as an early warning.
- Organise diagnosis and action - every alarm is assessed by an analyst and, where needed, leads to a work order in the CMMS. Record the outcome for root cause analysis.
- Review and adjust - compare warnings with what was found on disassembly and adjust limits and intervals.
Knowledge matters as much as technology. ISO 18436-2 defines four certification categories for vibration analysts: category I collects data on set routes, category II diagnoses common faults and sets alarm bands, and categories III and IV handle advanced analysis and programme design.
🔐 How do you secure condition monitoring in the OT network?
Online condition monitoring adds sensors, gateways and often a vendor cloud platform to the OT network. This creates a new path between OT and IT that must be secured as carefully as any other connection.
- Segmentation - place CM servers and gateways in their own zone in line with IEC 62443 and the Purdue Model, with access control between zones
- One-way traffic - let measurement data flow only from OT to IT, using a data diode where requirements are high
- Encryption - use OPC UA with the SignAndEncrypt security mode or MQTT with TLS; OPC UA only encrypts when configured to do so
- Wireless sensors - WirelessHART encrypts with AES-128 and a join key per network; manage those keys as strictly as passwords
- Vendor access - limit remote access for the CM vendor to what is necessary and log it
- Monitoring - forward gateway and edge device logs to a SIEM
- Physical protection - protect sensors and edge devices against tampering and unauthorised access
Two European rules apply. Organisations covered by NIS2 and the Dutch Cybersecurity Act must also manage the security risks of suppliers that analyse machine data remotely. And since 12 September 2025, the Data Act has given users of connected machines the right to access the data their equipment generates.
❓ Frequently asked questions
What is the difference between condition monitoring and predictive maintenance?
Condition monitoring measures and analyses the current condition of a machine, for example with vibration or temperature sensors. Predictive maintenance uses those measurements to forecast when a component will fail. Condition monitoring is therefore the foundation on which predictive maintenance is built.
Which standard applies to condition monitoring?
The general guideline for condition monitoring is ISO 17359:2018, which describes how to set up a monitoring programme. ISO 13374 covers data processing, ISO 20816 the evaluation of vibration and ISO 18436 the certification of analysts. These standards are not legally mandatory, but they are widely used internationally.
Is online condition monitoring always better than measurement routes?
Online condition monitoring is not always better; it costs more per measurement point and is mainly worthwhile for critical or hard-to-reach machines. For large numbers of less critical pumps and motors, periodic routes with a portable instrument are often more cost-effective. Many companies combine both forms.
Which machines are suitable for condition monitoring?
Condition monitoring is most worthwhile for critical machines with a measurable failure mode, such as pumps, motors, compressors, gearboxes and turbines. A criticality analysis determines which assets qualify first. For cheap or redundant components, corrective or preventive maintenance is often more economical.
Does condition monitoring need AI?
Condition monitoring also works without AI, using fixed alarm limits, trend analysis and the judgement of an experienced analyst. Machine learning adds value with thousands of measurement points or complex patterns, for example through anomaly detection. Start with simple, explainable methods.
What security risks does condition monitoring introduce?
Online condition monitoring adds sensors, gateways and cloud connections to the OT network, often with access for an external vendor. A poorly secured gateway can open an attack path to the control system. Segment the network, let data flow in one direction and monitor every connection.
📌 In summary
Condition monitoring measures and analyses the condition of machinery, so that developing faults become visible long before they cause a breakdown. With a set-up based on ISO 17359, the right mix of measurement routes and online monitoring, follow-up in the CMMS and a well-secured data flow between OT and IT, condition monitoring is a critical link in the transition to Smart Industry.