4 ML use cases which show that predictive maintenance is definitely here to stay


Jeroen Boeye

Head of Machine Learning

The maintenance of production line machinery and equipment, but also of other assets such as trucks and batteries represents a major expense. There is no single manufacturing or transport company that isn’t aware of this challenge. In addition, the reactive maintenance approach which is typically used today results in significant unplanned downtime. Although reactive maintenance is putting pressure on their cost position and causes disruptions in both their upstream and downstream value chain, manufacturers and transport companies don’t have to worry about these downsides because they are unavoidable, aren’t they?

No, they fortunately aren’t. After all, manufacturers and transport companies gather a vast amount of sensor data that can be leveraged to gain valuable insights about their production equipment and logistics assets. These insights are provided by Machine Learning techniques, making a shift towards predictive maintenance possible. In what follows, four use cases related to predictive maintenance will be discussed, which are highly relevant to the already transforming manufacturing and transport industries.

Predictive maintenance of failing components

When will a certain component fail? Initially this sounds like a question that you could only ask to soothsayers. Nowadays, however, we can ask Machine Learning algorithms to predict component failures. By monitoring the state of the equipment, Machine Learning will proactively identify the problem before it occurs. By doing this, it will also eliminate possible points of failures or degraded performance and thus enable an almost uninterrupted production process. No wonder manufacturers and transport companies are eager to leave the reactive maintenance approach and take the predictive one.

What is the impact of predictive maintenance of failing components on your business?

Even though the goal of this use case is quiet straight-forward, the positive effects of implementing it in your production or transportation process are various and huge. Surprisingly, the choice for predictive maintenance is for most manufacturers and transport companies a strategic decision. By minimizing downtime, you avoid production losses which result in unhappy customers and major costs. From a safety perspective predictive maintenance is very important as well. When Machine Learning is able to give you early alerts about possible future component failures, you can avoid critical incidents in your production and transportation process.

Anomaly detection for unexpected shutdowns

In the chemical industry, for example, unexpected shutdowns in chemical production processes (e.g. overheat) are most of the time caused by unusual events in production equipment such as kettles, tanks, pumps etc. That’s why chemical manufacturers want nothing more than identifying these extreme events. Anomaly detection is therefore deployed. Generally speaking, anomaly detection consists of Machine Learning techniques which are able to find outliers in vast amounts of sensor data. With a high precision, anomaly detection can tell you where an error occurs and subsequently inform the responsible employees/departments to act.

What is the impact of anomaly detection for unexpected shutdowns on your business?

As you know, production staff in manufacturing processes has to keep an eye on hundreds or thousands of production variables. It’s almost impossible for them – as for every human-being – to notice every outlier in the sensor data. Let alone to assess whether a certain data point is extreme or not. By contrast, Machine Learning can do it very quickly and accurately. As a result, you will avoid the Butter Effect by which a small change (in this case a small deviation in your production process) can have a huge impact (in this case an unexpected shutdown and so big production losses and increasing costs).

Better maintenance backlog prioritization

For most manufacturers and transport companies, a maintenance backlog feels like a never-ending to do list. In order to avoid safety issues, breakdowns or damage to production or logistics equipment, there are just so many checks and repairs that have to be executed. Obviously, this makes it complicated for manufacturers and transport companies to prioritize maintenance tasks. After all, you don’t want to waste your time and money on irrelevant maintenance. Fortunately, Machine Learning can prioritize the urgency and potential impact of maintenance by looking through maintenance and incident logs.

What is the impact of a better maintenance backlog prioritization on your business?

The main benefit of a better maintenance backlog prioritization is that your technicians perform the most urgent and critical maintenance tasks first. Normally, manufacturers and transport companies experience a vicious cycle when it comes to their maintenance backlogs: 1) they don’t know exactly which maintenance tasks to perform first, 2) as a result, they execute some non-urgent tasks and don’t execute urgent tasks, 3) because they have ignored some urgent maintenance tasks, a breakdown in the production process occurs, which results in new items on the never-ending to do list called the maintenance backlog. Machine Learning is able to break this vicious cycle for you. This will save you a lot of costs and valuable resources.

Improve asset efficiency

There is no doubt that most manufacturing and transport industries are capital-intensive. The production processes in a typical manufacturing plant, for instance, heavily rely on the operating machines. As a consequence, the manufacturers’ profit heavily rely on the operating machines’ performance. The same applies to transport companies who heavily rely on their logistics assets such as trucks. That’s why manufacturers and transport companies are eager to use predictive maintenance to decide when to deploy the production and transportation equipment and when to turn it off for reparation. Leveraging sensor data from production and transportation equipment, Machine Learning provides these insights and gives you the opportunity to optimize asset efficiency.

What is the impact of an improved asset efficiency on your business?

Capital investments are a major cost position for manufacturers and transport companies, so it’s essential to them to receive the highest possible return on these investments. By implementing Machine Learning to obtain predictive maintenance, you are able to deploy your assets in the most efficient way. As a result, you will save a lot of maintenance costs and efforts, and moreover expand the lifetime of your production assets. Last but not least, the Machine Learning-based prediction model will generate alerts and recommendations to enhance your production and transportation assets’ performance, which will result in more operating profit.


Although we have shown you that predictive maintenance has a huge potential in the manufacturing and transportation industry, we realize that the world of AI and Machine Learning could be overwhelming for many. Fortunately, Faktion can be your reliable compass to guide you through the AI landscape. Our successful people-projects-processes-approach will help your manufacturing or transportation firm through its digital transformation towards smart manufacturing or logistics. Get in touch with us today!

Get in touch!

Inquiry for your POC

Scroll to Top