Predictive maintenance: A look into the crystal ball

At Merck KGaA, Darmstadt, Germany, we are already using predictive maintenance in our production.

It’s maintenance time!

Once every six months or after 1,000 parts are produced – maintenance on systems and machines is still carried out in many companies in accordance with a fixed schedule. It doesn’t matter whether maintenance is actually necessary. However, this non-needs-based approach could soon be history. After all, the development of constantly improving digital technologies based on increasingly smart algorithms is bringing a still relatively new solution to the fore: predictive maintenance.

Predictive maintenance works by constantly collecting process and machine data and analyzing them in real time. For example, sensors measure the wear of the interior of the machines and the power supply as well as environmental factors, such as temperature and humidity. Based on the data, algorithms predict when a machine might show defects or even completely malfunction. In this way, maintenance can be planned in a much more targeted and needs-based manner.

Use in practice

Predictive maintenance has been assigned a key role in Industry 4.0. An increasing number of companies are already using this technology. For example, at its site in Bremen, German carmaker Daimler uses predictive maintenance in the manufacture of the EQC electric car. It uses production data to make predictions about potential defects in the systems.

Deutsche Bahn already launched a pilot project for smart maintenance of its tracks in 2016 and equipped 28,000 switches with sensors. The so-called “switch ECG” measures the power applied to the switch drive motors and monitors the temperature and material of the switches. Deviations from the normal values are recognized as early as possible. According to Deutsche Bahn’s own information, this prevented 3,600 malfunctions in 2019.

At Merck KGaA, Darmstadt, Germany, we too have already had the topic of predictive maintenance on our radar for some time. The possibilities of digitalization now enable us to use increasingly smart solutions. For example, as part of a pilot project at our sites in Darmstadt and Gernsheim, we equipped centrifuges, which are used to purify liquid crystals or effect pigments, with the corresponding technology. In the interior of the machines, numerous sensors measure the tiniest vibrations or changes in the power supply – both of which are signs of wear and tear of the machine. Then, using an Internet of Things (IoT) adapter, the data is sent to the central network of Merck KGaA, Darmstadt, Germany via a radio signal, where it is analyzed, visualized, and used for targeted maintenance planning.

To further advance development in the area of predictive maintenance, we also recently started collaborating with the Israeli start-up Feelit. The start-up has developed a sticker sensor with printed nanotechnology, which detects structural changes in mechanical systems and thus detects malfunctions.

This precise maintenance additionally increases the lifespan of systems because unnecessary production stops are avoided, and signs of wear and tear are detected in their early stages.

According to a study by the consulting firm Deloitte, companies can increase their productivity by 25% on average through predictive maintenance. Costs can be reduced by up to 25% and malfunctions by up to 70%. According to the German Industry 4.0 index, between a quarter and a third of industrial companies now use predictive maintenance technologies in day-to-day operations.

The challenge of digitalization

However, there are also reasons why predictive maintenance is not yet more widespread, despite the undeniable advantages of the technology. According to a study by the consulting firm Pierre Audoin Consultants, one of the central challenges in implementing this technology is the handling of large volumes of data. After all, it does require an appropriately powerful infrastructure for transmitting, processing, and storing data.

Another reason why predictive maintenance is still in its infancy in many places is simply the lack of trained personnel. This example shows once again how crucial it is to include digital skills as a core element of school curricula and advanced training programs within companies. That’s why I am happy to advocate for this topic regularly, both within our industry and within my role at BAVC.

In the long term, there will be no getting around predictive maintenance. That’s why it’s all the more important to continuously develop this technology further. After all, it has a great deal of potential, not only in industrial production but also anywhere where technology is used. As a result, predictive maintenance could also increasingly be used in the future to maintain cars or building technology according to need. In any case, at Merck KGaA, Darmstadt, Germany, we are already using the potential of predictive maintenance and will continue to contribute to the development of this technology.