The main challenge in maintenance at many companies: The team is permanently occupied with eliminating the consequences of unplanned breakdowns as quickly as possible so that production can continue and the breakdown does not have too great an impact on subsequent processes. A further challenge in maintenance lies in personnel management when the knowledge is concentrated in a single person as a specialist. If this person is absent due to illness or vacation or leaves the company, this creates a large knowledge gap that cannot be easily filled.

Companies often only become interested in optimizing maintenance after a serious breakdown has occurred that has caused major damage. Only then are resources made available to prevent such an event from happening again. Some efforts to improve maintenance are not effective due to a lack of resources. In other cases, requirements such as the implementation of predictive maintenance are passed down from the corporate level, but the local team needs guidance and support to implement them. Still other companies are overwhelmed by the selection of offers and terminology, they need orientation and a sensible approach.

Optimize maintenance management

In order to improve their maintenance management, companies need a robust maintenance plan. It is based on four pillars: failure-oriented maintenance, in which defective parts are replaced, and runtime-related replacement, i.e. preventive or time-based maintenance (preventive maintenance). The third pillar is predictive maintenance: Replacement here is not blind, but targeted, based on measurement data. The last pillar is proactive maintenance, in which optimum environmental conditions are created for the machines in order to eliminate sources of error. If a good data basis is available, the bridge to prescriptive maintenance can also be built here.

The optimal design of the maintenance plan combines these options into a perfect mix. The use of a condition-based maintenance strategy in particular can significantly increase plant availability. Unplanned downtime can be largely avoided and investments can be made in the right places. Condition-based maintenance makes sense for expensive systems or system components with long delivery times, for example if a gear element cannot be delivered for another six months. Systems that are used for the production of high-priced products such as pharmaceuticals should also be monitored.

But even seemingly simple components such as a good lubrication program, which is more akin to preventive maintenance, are important components of an optimum maintenance plan. Because you must not forget: The majority of unplanned breakdowns occur due to incorrect lubrication, when too much, too little or the wrong product is used. This can be prevented. And you may be surprised how important good data collection is here. Basically, a solid database is an important prerequisite for the continuous optimization of a maintenance plan. It needs to be collected, connected, managed and analyzed centrally on a platform. Algorithms, analyses and augmented intelligence can then be used to detect damage at an early stage and identify recurring problems.

The pitfalls on the way to optimal maintenance

There are a few pitfalls on the path to optimized maintenance, as companies often find it difficult to find the right mix. For example, preventive replacement based on service life does not make sense per se. This often involves replacing parts that are still functional. What's more, the probability of problems is statistically higher after an installation, while the majority of failures are statistically independent of the service life.
Some companies use their budget incorrectly or do not know which systems are really critical. For example, they have a vibration measurement carried out on their systems once a year. It would make more sense to monitor the most important systems constantly rather than randomly.

Another problem is unsuitable sensor technology: The right sensor technology can only be selected and installed if you know which cases of damage are to be detected. It is the first building block for accurate and good data: If a vibration sensor measures a frequency range that is too low or too high, for example, faults outside the measurement window will not be detected and valuable reaction time may be lost. Relevant knowledge about the parameterization of the sensor technology is essential here; in addition, the assignment of data stems and measuring points must be correct. Further problems arise if measurements are taken at the wrong time or not often enough. However, the main hurdle for many companies is the lack of a central data platform. This is because data must be networked in order to create added value.

Draw up the maintenance plan

Due to these various challenges, it can be useful to obtain external support for drawing up the maintenance plan. For example, an audit or assessment can first be carried out and the current situation recorded on site. The goal is then determined: The company sets the benchmark, depending on whether it wants to compare itself with another plant or is aiming for an industry or even world-class standard.

It should then be shown which measures can be used to achieve the quickest success. The criticality analysis is central to the process. This breaks down which systems are particularly critical for production and whose failure must be avoided at all costs: permanent monitoring is recommended for these. For other machines, the focus can be on allowing damage to occur and rectifying it as quickly as possible.

The implementation options can then be determined in cooperation with the respective company: It can implement the measures of the recommendations for action itself, be guided or outsource the process. Implementation is followed by monitoring and measuring the systems. With comprehensive diagnostics such as infrared, vibration measurements or ultrasound, faults can be detected at an early stage so that they can be rectified in a planned manner.

Central data platform for prescriptive maintenance

Data is the focus of Industry 4.0 and is therefore the be-all and end-all for production. With a powerful data platform, it can be analyzed and important insights can be gained from it. Generated KPIs (key performance indicators) can then serve as a basis for management decisions.

A data platform that integrates, consolidates and provides all collected data and reports must therefore be at the center of the cycle from recording the status quo to implementing measures, monitoring and readjustment. The platform should be characterized by its openness to other systems and sensors in order to be able to draw data from other systems.

Such a platform then enables the step from predictive to prescriptive maintenance, from predictive maintenance to an agile strategy that constantly optimizes, increases performance and further minimizes risks. To achieve this, data from various sources, devices, sensors and history are brought together - with standardized connectivity and IoT protocols. It is enriched and processed from the perspective of big data and machine learning. New insights can be gained from this to continuously optimize maintenance.

This is because predictive maintenance does not prevent damage from occurring in a system. Condition-based maintenance only helps to detect developing damage at an early stage and thus avoid unplanned downtime. Prescriptive maintenance, on the other hand, asks what needs to be done to prevent the damage from occurring in the first place. And ideally, the machine should be able to answer this question itself. But until that happens, humans have to take action and initiate appropriate measures. Regardless of this, a good data basis is crucial for the success of the measures taken.

Effective maintenance starts with a plan that integrates condition-based technologies to prevent unplanned downtime of production equipment. With the right mix of maintenance approaches, companies are able to act proactively instead of being stuck reacting. The prerequisite for this is a solid database from monitoring the systems with suitable sensor technology. If this data is consolidated on a central platform, the step towards prescriptive maintenance can be taken.