The most common maintenance method today is preventive maintenance.

Wear parts are replaced after certain periods in order to prevent unplanned downtime.

And maintenance work is carried out, such as oil changes, toothed belt changes or simply lubrication work.

Wear parts are replaced well before the end of their service life, which leads to avoidable costs. But an unplanned outage would cause even higher costs.

Wouldn't it be great if you had a crystal ball that could predict exactly when something would break?

No, this crystal ball has not yet been invented and will probably never exist.

But there is something that can be compared to her.

A software that can use AI algorithms and existing sensor data to predict when a component will soon fail.

Sounds great, but is there anything else that needs to be in place for the software to work?

Of course there is something else, that is data. And a lot of them are needed. And how do I get it?

Modern machines are packed with sensors. And the data from the sensors must be entered into a database in real time. These can be transmitted using LNA cables. But what means a lot of work to lay the cables from each machine. Or using WiFi, but it has to be a very fast WiFi, or by radio with LoRa. Or with 5G.

To understand this better, a small example:

A sensor monitors the temperature of a ball bearing. This is written to the database at exactly the right time.

Another sensor monitors the vibration of the ball bearing. These are also written to the database at exactly the right time.

All good things come in threes, a noise sensor monitors the noise of the ball bearing. These are also written to the database.

By the way, there are sensors that can monitor all 3 values.

If the ball bearing is working optimally, then there is a certain pattern in the database. And as long as the ball bearing works well, this pattern will always be the same.

Eventually, wear will build up, so there may be vibration, or the noise level changes, or the bearing just gets warmer than usual.

Maintenance now receives information from the software that something is wrong with ball bearing 5. So a maintenance employee goes to the machine operator and asks him if he noticed anything. There can come, for example, yes it has become a little louder.

So both look for something to be seen or felt. Yes, you can tell that something is going on. So the maintenance engineer orders a new ball bearing and then installs it when the machine is not needed.

So there is no unplanned downtime, and there is no replacement too early.

So activities only have to be started when something reaches its wear limit.

Would you like to learn more about it? Then click here