This makes it possible to plan machine downtimes, organize the necessary tools and spare parts at an early stage and instruct personnel in good time. But why doesn't every company use predictive maintenance?

Unfortunately, error events in highly complex technical systems are not as predictable as one would hope - at most one in ten potential causes of error can be predicted well to very well. This is the perfect candidate for the predictive maintenance strategy. All others have too low a degree of predictability. A typical example of this is one of the most frequent fault events in highly networked systems: line faults. Causes can include a short circuit, a break or an open plug. These causes neither announce themselves nor can they be predicted in any other way. However, as they also need to be detected and rectified, adapted maintenance strategies are required.

What are technical system events?

In the case of an event, a defined cause of failure has an effect and influences the normal functioning of the technical system. In order to better understand the causes of faults in technical systems, the resulting events can be roughly divided into categories with common characteristics.

• Environmental events have their causes in the surroundings around the technical system or even in the environment. They are usually exceedances of limit values of physical quantities. Examples can be too high a temperature, too high a humidity, too high an air pressure or too many dust particles.
• Process events are errors whose causes lie in the product life cycle or in its actual implementation. This can be due to missing settings or programming of the technical system, as well as missing work equipment or incorrect sequences in the application of the production process.
• Error events are caused by errors in the system. This includes the classic causes of errors. Defective control units, jammed actuators or pressure loss are examples in this category.
• Functional events are caused by the implementation of a function of the technical system and lead to functional restrictions. Examples of possible causes for a functional event are the activation or deactivation of special system functions, functional safety mechanisms and exceeding of counter values, control limits or results of system calculations.
• Communication events are caused by networking with other systems. These events can occur through notifications, warnings, information or messages from devices outside the system. However, external access and interventions by diagnostic functions are also assigned here. Examples include incorrect activation codes, event messages via the Internet or the execution of an update.

Fingerprint of events of technical systems.

© Synostics

Each event in each of the five categories leads to a disruption in the normal functioning of the technical system. In any case, it is important for the productivity of the machine or system concerned to eliminate the fault quickly and in a targeted manner or to react quickly to the event.

Even at first glance at the individual categories, however, it is clear that there are causes of faults that can be predicted with a little experience. One example of this is the classic wear and tear of parts. On the other hand, there are causes of failure that cannot be predicted. Communication or process events are usually outside the system under consideration and therefore cannot be predicted.

Obviously, different maintenance strategies can be used in the above cases. In order to find out which maintenance strategy can be used for which event, it is necessary to take a more differentiated look at the individual events.

The fingerprint of technical events

Each event of a technical system has specific characteristics and properties that can be categorized and measured and considered as a kind of fingerprint of the fault event. The following are examples of characteristics that have been used for years in the automotive and mechanical engineering industries. These examples do not claim to be exhaustive; depending on the industry and system, more or fewer criteria may be taken into account:

The best-known maintenance strategies for technical fault events.

© Synostics

• Criticality describes the severity of an event in relation to the probability or frequency of its occurrence. The higher the value, the more likely the event is, the more serious its effects are and the faster the affected function in the system is interrupted.
• The identity describes how well localization and clear identification can be carried out in the event of a failure or event. The higher the value, the easier it is to detect and find the event.
• The reparability describes the recoverability after a failure or event in connection with the effort, costs and duration required to rectify the event. The higher the value, the easier it is to rectify an event.
• Predictability describes the predictability of an event in connection with its failure measurability, the effort required for analysis and the ability to recognize patterns. The higher the value, the better an event can be predicted.
• Preventability describes the avoidability of an event in relation to its interchangeability, its improvability and the effort required to avoid it. The higher the value, the easier it is to avoid an event.
• Intensity is the average of all previous characteristics and allows the event to be weighted in comparison to other events.
• Vitality indicates the stage in the life of the technical system at which an event can occur. Compared to the other characteristics, it does not represent a single value, but a time range.

When comparing the characteristics and properties of different events with each other, it turns out that a typical fingerprint is formed for each event. This fingerprint provides an initial indication of possible maintenance strategies for the corresponding event. Similar fingerprints require the same maintenance strategies.

The best-known maintenance strategies

A maintenance strategy describes the procedure for maintaining or even increasing the productivity and life expectancy of a system. There are a number of different strategies that differ in terms of when they are carried out relative to an event and the individual tasks within the strategy implementation.

• Reactive maintenance is corrective maintenance that is carried out without delay after the fault is detected in order to avoid unacceptable consequences. All events that can be easily identified and localized are very suitable for reactive maintenance. On the other hand, events that have little or no functional impact and are very unlikely to occur do not need to be considered for this strategy. Safety-relevant faults should be rectified using other maintenance strategies.
  The advantages of this maintenance strategy are the maximum utilization of the service life of the components, the low planning effort, the ease of implementation and the involvement of the users in the maintenance control. The duration is only determined by the repair itself. However, unplanned activities to be carried out immediately, the risk of longer downtimes and lower system availability must be taken into account as disadvantages.
• Deferred maintenance or active maintenance is corrective maintenance that is not carried out immediately after a fault is detected, but is deferred according to predefined maintenance rules. It is a secondary form of deferred maintenance with the special feature that the time of repair is planned. All fault events that can be easily identified and localized are suitable. Safety-relevant faults and functional restrictions must be rectified using other maintenance strategies. However, faults that can be rectified quickly and immediately do not make sense for this strategy.
  This maintenance strategy has the same advantages and disadvantages as the failing maintenance strategy. However, the additional planning effort and the risk of consequential damage should also be mentioned as disadvantages.
• Predetermined maintenance or preventive maintenance is preventive maintenance that is carried out according to a fixed schedule or after a fixed number of units of use, but without prior condition assessment. This is classic maintenance. All events that cause low costs and for which maintenance does not take too long in relation to a possible repair are suitable for this strategy. Failure events that are difficult to repair and unlikely events that have no impact on functional safety and operation are not suitable.
  The advantages of this strategy are the ability to plan the work (deadline and duration), precise knowledge of the processes and the required spare parts, high material availability, high work quality, exact calculability, the reduction of unplanned failures and the increased service life of the components. On the other hand, increased costs and additional working hours for planning and carrying out maintenance and a lack of knowledge about the actual failure behavior of the system are disadvantageous.
• Predictive maintenance is preventive maintenance that is carried out on the basis of a prediction. This prediction is based on known properties or experience from past analyses or the occurrence of important parameters that characterize the degradation of an element. This maintenance strategy is used to prevent the future malfunction of a system. The components of a potential failure event are replaced or the effects of the causes of failure are reduced before the event occurs. The important thing here is that the event must have a certain degree of predictability.
  The advantages are that the work can be planned, the processes and required spare parts are known, high material availability, high work quality, exact calculability, increased system availability, fewer unplanned breakdowns and an increased service life of the components. Disadvantages, on the other hand, are the need for sufficient lead time for planning, intelligent algorithms and sufficient data sets for pattern recognition. Other disadvantages include higher effort and higher costs due to the planning and implementation of increased maintenance work, high demands on personnel qualifications for the development of algorithms and the need for knowledge about the failure behavior of the system. Furthermore, this strategy can only be implemented well after the start of production, as the necessary experience must first be gathered. Due to the large number of disadvantages, both unlikely and rare events with little impact on functional safety and operation as well as difficult-to-repair and cost-intensive events are not suitable for predictive maintenance.

Event fingerprint masking: The graphic shows the masks of the various maintenance strategies as an example, as well as an example fingerprint of an event and how it fits into the respective masks. The selected event fits very well into both the mask of the failing and the displacing maintenance, while the other maintenance strategies would be a rather poor choice.

© Synostics

• Condition-based maintenance is condition-oriented maintenance. Possible maintenance measures result from assessments of the physical condition and, if necessary, further analyses. It is a predictive, knowledge-based strategy. All events with a certain degree of predictability and good localization are suitable. The advantages and disadvantages are identical to predictive maintenance.
• Suppressive maintenance is a strategy in which a recognized or existing event is ignored. It is not real maintenance, but rather a solution for dealing with specific fault events. With this maintenance strategy, a known or unknown system malfunction is not rectified.
  This strategy only makes sense for events with little or no functional impact. The advantage is that no costs are incurred. The disadvantage, however, is that at the very least functional restrictions of the system are to be expected, and there may even be an increased risk of failure because no repair is carried out.

Suitable for every event

The author: Heino Brose is Managing Director of Synostik in Oebisfelde.

© Synostics

The most effective maintenance is achieved when the maintenance strategy used matches the maintenance event in question. The decisive task is therefore to assign the usable maintenance strategies to an event. How can this be achieved?

Event fingerprint masking is a simple and very successful method. This method compares the properties of the maintenance strategies with the fingerprints of technical events.

In principle, predictive maintenance is particularly suitable for events with a high level of predictability. However, as this maintenance strategy also has serious disadvantages - especially economic ones - additional specific characteristics should be considered.

In the case of easily identifiable or easily repairable events, it may be better to simply wait until the event occurs and then deal with it. For easily preventable events, it is probably also better to prevent the event from occurring rather than replacing complete components early on, for example. However, if the fingerprint of the event shows low preventability, low identity and low reparability in addition to high predictability, predictive maintenance may be a good choice.

This approach results in special masking for each individual strategy, which can be used for the analysis and strategy decision of each individual event. A maintenance strategy is well applicable for an event if the typical fingerprint of the event fits into the mask of the maintenance strategy.

If the industry makes use of existing experience with events from other sectors, perfectly adapted diagnostic strategies can provide an economic advantage in maintenance.

Video tip on the topic