If a roller bearing or gearbox breaks down unexpectedly, the consequences can be devastating. Not only that a whole range of costs suddenly arise that ideally could have been avoided. There is also the issue of delivery reliability, with everything that goes with it. Therefore, there is already a whole range of monitoring options or even early warning systems that aim to avoid or prevent exactly that. The differences between them are serious. We therefore wanted to know more and asked an expert.
Monitoring systems offer very different services. What helps in the decision-making process?
Markus Loinig: When it comes to comparing different providers in order to come to a decision, you have to ask exactly. There are indeed many systems that offer monitoring of plant components. Customers like to use "established" or "conventional" systems – but what exactly is meant by the terms "established" or "conventional"? When I ask this question, people usually shrug their shoulders. This is because the market for surveillance systems is unbelievably large and at the same time confusing. These terms are often used when people want to rely on familiar things, such as vibration technology. Everyone knows it, it seems reliable. But in the end, it is important to know what the systems really do, where exactly their focus is. It is therefore helpful to deal with the most important questions, which at the same time make the differences clear and provide clarity.
What kind of questions can these be? What questions should you ask yourself?
Markus Loinig: For example, it is important to know what exactly the system monitors and what happens when damage is registered. Is it displayed what and how badly something is damaged? Does the system trigger an alarm as soon as a roller bearing is damaged, for example? And is this alarm really correct and necessary at all times? What are the criteria for an alarm in the first place? Or also the topic of remaining service life: does the system calculate how long it will last and safely perform its task? And if so: how and with what technology? Is a maintenance order automatically triggered by the monitoring? Is it possibly additionally monitored whether and how a roller bearing is lubricated? How does the system deal with a false alarm? Is it capable of learning? As you can see, these initial questions alone open up the possibilities that already exist on the market today.
"Conventional" or "established" systems often mean technology based on the detection of vibrations. Is this the best way?
Markus Loinig: Vibration technology is based on experience and, consequently, on a fundamental assumption: if there is a threat of roller bearing damage, if a gearbox breaks down, then it starts to vibrate more, so as a rule the vibration value increases. A threshold value is exceeded. That is how I know that the gearbox or the roller bearing is in danger of breaking down. This can be easily determined or measured. There is only one problem that is usually overlooked: this assumption does not work the other way round. Conversely, an increase in vibrations does not necessarily mean that there is a risk of bearing damage. In fact, there are several causes for vibrations and, since they have the property of spreading out and gladly cover long distances, one must consequently always expect a false alarm.
What exactly causes a false alarm, for example?
Markus Loinig: The situation is as simple as it is understandable: if several plants are next to each other in a factory, vibrations overlap, for example. Maybe a forklift driver just drives by? Do operating conditions change, such as speeds or changing loads on roller bearings, or are there other sources of interference such as imbalances or alignment problems? All this causes vibration values to rise and causes vibration sensors to suddenly detect increased vibration values. Quite normal, is not it? So yes, there are a lot of different sources of vibration. But they have nothing whatsoever to do with deterioration of your component.
But we are talking about an "established" technology here – tried and tested. Basically, everything works, does not it?
Markus Loinig: The short answer is: yes, this technology works – but really reliably only where the environmental situation is clearly defined and simply designed. In this case, "simple" means that an increase in vibrations can clearly indicate a deterioration of the roller bearing or gear.
So where do these technologies really work well?
Markus Loinig: As a rule, where the components to be monitored, such as the roller bearing or the gearbox, are maximally isolated from other, parallel vibration sources and operate at almost constant speed. Constant speed because different speeds also cause different vibration values. So – if you have these special, stable framework conditions, in reality you will come across many companies that have had good experiences with "conventional" monitoring systems and again others that have had bad experiences. These "good" or "bad" experiences mean translated: The technology "works" almost smoothly where it is used in simple, stable environments. And it just does not "work" under complex conditions – this technology cannot do a good job because you cannot reliably detect a damaged roller bearing, for example.
What technologies could you rely on if you do not have simple, stable environmental conditions?
Markus Loinig: The issues just described have led to the fact that NASA, for example, has been using a different technology for decades to monitor its roller bearings. It relies on the detection of ultrasonic frequencies or acoustic emissions. A completely different approach. A study has shown that faults in roller bearings cannot be detected earlier with any other technology. Technological progress has made it possible to use this technology on a broad scale. Sensor technology has become affordable. And if you now combine this technology with innovations in the field of artificial intelligence (AI), the result is unbeatably advantageous application possibilities. We are no longer just talking about condition monitoring, but rather predictive maintenance in its best form.
What exactly does predictive maintenance do?
Markus Loinig: Predictive maintenance not only provides a precise view of the current situation. This technology can also look into the future. For example, it answers the question of how long a roller bearing will continue to fulfil its task. The topic of remaining service life mentioned earlier. Predictive maintenance provides reliable answers here. This saves costs, because the product life of such a bearing is utilised to the maximum and, above all, safely. As a company, you can react in time and avoid failures. So you are on the safe side. A clear recommendation, I would say.
