Condition monitoring: condition-based maintenance & predictive maintenance

Condition monitoring is a requirement for modern maintenance measures in industrial processes

Especially in the age of Industry 4.0, it is more important than ever to rely on smart systems that enable proactive action and contribute to increasing production efficiency.

To ensure that the terms "condition monitoring", "condition-based maintenance" and "predictive maintenance" don't just remain abstract buzzwords for you, this Automation24 guide explains how these concepts work in concrete terms, how you can benefit from them and why you shouldn't wait any longer to optimise your industrial processes and take them to the next level.

What is condition monitoring?

Condition monitoring covers the condition monitoring of machines and systems and is a fundamental pillar of Industry 4.0.

In this context, the so-called "digital shadow" is used, through which the current state of a system can be virtually represented. The digital shadow refers to process data generated by machines while they are in operation. These can incluse physical parameters such as temperature, vibration, oscillation, pressure, humidity and electrical signals or even the oil quality. This allows the condition of these machines to be checked and analysed at any time, typically in real-time or at predetermined intervals.

This enables early detection of potentially critical conditions and the taking of appropriate measures, thus systematically preventing failures and downtimes. Condition monitoring is therefore a central element of modern maintenance strategies.

What components does condition monitoring consist of?

All the essential components for solid condition monitoring are most likely already in your systems today. This can look like this, for example:

1. Data recording: Sensors record relevant machine and process data. These can be the following, for example:

• Vibration sensor: Warns if machine vibrations deviate from the standards.
• Inductive sensor: Detects changes in the distances between components using the sensing range for early bearing wear detection.
• Ultrasonic sensor: Indicates errors in the process if there is no sound reflection.

2. Data transmission: IO-Link and similar protocols ensure continuous communication.

3. Programmable logic controller (PLC): Provides the sensor data to the network.

What advantages does condition monitoring offer?

Efficient condition monitoring offers you numerous benefits for your business, including

• Early detection of machine problems: Damage or wear and tear are recognised before they lead to expensive breakdowns.
• Preventive elimination of the cause: A defective part can be repaired or replaced before damage occurs.
• Wear reduction: Proactive intervention also prevents wear on other machine parts.
• Increased efficiency: Optimised maintenance intervals avoid unnecessary maintenance work and increase machine availability.
• Cost reduction: Targeted measures minimise expensive repairs and production downtime and reduce operating and maintenance costs.
• Increased safety: Monitoring sensitive system components reduces the risk of accidents.

Practical overview of common parameters for condition monitoring

What is predictive maintenance (PdM)?

While condition-based monitoring (CBM) already provides an efficient basis for monitoring the condition of the machine, predictive maintenance (PdM) goes one step further and uses the data collected by CBM in combination with more advanced analysis methods to accurately predict the optimum time for maintenance.

"Predictive maintenance" (PdM) aims to predict when maintenance will be required - before an acute malfunction occurs and the industrial process comes to a standstill.

With periodic maintenance, maintenance is carried out at fixed times regardless of the system status, which usually results in unnecessary downtime. While reactive maintenance addresses issues only after critical deviations or impairments occur, predictive maintenance acts earlier, proactively preventing anomalies and failures.

Compared to traditional maintenance strategies, predictive maintenance focuses on preventive rather than periodic or reactive maintenance measures.

What advantages does predictive maintenance offer?

As predictive maintenance is an advanced version of modern maintenance strategies, a separate chapter is dedicated to the numerous short-term and long-term benefits.

• Optimised maintenance management: Time, a valuable resource, is now used exclusively for necessary, plannable maintenance intervals - unexpected faults are eliminated. If a sensor is defective, it can even be replaced during operation thanks to modern communication technologies such as iO-Link, provided an identical sensor takes its place.
• Maximising production efficiency and machine performance: Early identification of declining machine performance and proactive maintenance optimises the performance and service life of the machines and minimises downtimes. Plant availability is increased and the process result is consistently maintained at a high level of quality.
• Increased efficiency: Maintenance costs are no longer incurred in the event of sudden malfunctions, but only for planned and cost-efficient maintenance. Unnecessary and costly process downtimes are a thing of the past.
• Increased safety: Early repair of defective system components prevents potentially dangerous machine faults, thereby preventing safety-critical situations and protecting both machines and applications.

The parameter data is transmitted to local edge computing devices or to centralised cloud-based systems, where it is collected and processed using innovative analysis tools such as moneo. This allows failures to be predicted at an early stage and appropriate maintenance measures to be planned.

By utilising big data, machine learning and artificial intelligence (AI), the analysis tools continuously "learn" from the collected data and deliver increasingly precise forecasts. With the help of advanced data modelling, patterns and correlations are derived from these large amounts of data over time, making it possible to identify failures and maintenance requirements at an early stage. As a result, the failure probabilities initially calculated by the algorithm adapt to changing conditions and are recalculated accordingly. In this way , the predictions are always based on the actual prevailing conditions.

Users can also actively contribute to "teaching" the analysis tool by enriching the analysis and forecasting models with annotated data and specific feedback. Annotated data contains important additional information, such as downtimes or operating states, which help to further increase the accuracy of the model.

In order for maintenance measures to be truly "predictive", it is imperative that the data is transmitted in real time. This is where advanced technologies such as IO-Link come into play, enabling seamless communication between sensors, actuators and controllers. These ensure that all relevant data is processed quickly and reliably so that maintenance can be carried out before faults even occur.

In addition to failure prediction, which is based on the parameters processed by analysis tools, "digital twins" can also be used - virtual images or models of machines and systems that simulate their condition and functionality in real time.

The use of a Computerised Maintenance Management System (CMMS) is suitable for managing and planning the maintenance measures derived from the forecasts and simulations. With this system, maintenance actions can be efficiently planned, prioritised and documented, optimising the handling and transparency of the entire maintenance process.

Difference between condition monitoring, condition-based maintenance and predictive maintenance

Many wonder about the differences between "Condition monitoring", "Condition-based maintenance" and "Predictive maintenance".

"Condition monitoring is purely the monitoring of the condition of machines. The respective maintenance strategies "condition-based maintenance" and "predictive maintenance" then utilise this data. The strategies differ depending on whether maintenance is subsequently planned based on this data (CBM) or whether the data is processed by software, algorithms or AI. The aim of both approaches is to react before a problem occurs.

To illustrate this, you will find a comparison of the two maintenance strategies below:

How do you use condition monitoring?

In order to ensure smooth monitoring of your machines and systems, a number of technical requirements are necessary, but in many cases these can be easily implemented. Often, components that have already been installed already have the required functionalities and are simply not being utilised accordingly. If this is not the case, Automation24 explains the technical basics of condition monitoring right from the start.

What are the technical basics of condition monitoring?

1. Sensors: Every condition monitoring solution is based on high-quality sensors that measure key parameters such as temperature, vibration, pressure or humidity in real time. For example, vibration sensors can detect minimal deviations from target values that indicate imbalances, bearing problems or the onset of material fatigue. These are supplemented by temperature sensors. These recognise sudden overheating and damage to motors or gearboxes and can therefore prevent them.
2. Data analysis: The recorded data is transferred directly to a central data analysis platform, where it is analysed using modern algorithms. These algorithms identify patterns and anomalies that would be barely recognisable to a human observer. This creates a comprehensive picture of the machine's condition, which enables predictive maintenance and minimises unexpected failures.

IoT, digitalisation and real-time data transmission

Digitalisation has taken condition monitoring to a whole new level. Thanks to the integration of the Internet of Things (IoT), data from various sources can be merged and analysed in real time. One example: IoT-enabled vibration sensors are used to transmit measured values directly to a central control unit via a local network or the cloud. There, the data is immediately available for further processing.

Real-time data transmission is a key aspect of modern condition monitoring systems. It ensures that machine operators are informed about the current status of their systems at all times. To achieve this, technologies such as 5G networks and LPWAN (Low Power Wide Area Networks) are used, which enable reliable communication even with a large number of sensors and over long distances.

Data analysis: how intelligent monitoring works

Data analysis is the centrepiece of a condition monitoring system. This is where all the information collected by the sensors comes together. But how exactly does this process work?

Firstly, the data is recorded and pre-processed. Raw data is filtered and normalised to ensure that it is suitable for further analysis. Vibration sensors, for example, can deliver thousands of measurement points per second. Pre-processing removes irrelevant data such as ambient noise so that only the truly relevant information is forwarded.

In the next step, the cleansed data is analysed using artificial intelligence (AI) and machine learning (ML). Algorithms identify patterns that indicate a potential error. For example, an AI-supported model could recognise that a continuously increasing vibration indicates a bearing defect - even weeks before the defect actually occurs! Predictive maintenance thus provides you with reliable support and protects you from machine failures a lengthy time in advance.