How do we Use Online Oil Sensor Data in a Condition Monitoring Program?

Rafe sits down with Dr Guillermo Miro – Application Engineer at Atten2 to discuss all things regarding online/inline oil sampling and analysis. He explains that the the current online sensors do not measure data in the same way as standard lab analysis, so online data must be correlated with lab data to build algorithms. Once these algorithms are established, sites can use the online data to trend the degradation of the lubricant in real time and take more precise actions – whether that be oil change, top up, or filtration.

Rafe Britton: Current oil analysis is typically monitoring parameters such as oxidation. And as that oxidation number increases we watch the trend. For practical purposes, most businesses are just waiting until the oxidation number exceeds some value, and then they change the oil. Those values are usually determined by the OEM and they’re typically attached to a warranty or something like that.

How are we supposed to use the data coming from these new online sensors? Is it the same? Are we going to place limits on data or relying on trend analysis? How should companies and people be thinking of this data and integrating it into their current practices? 

Guillermo Miro: Your question is very good. I mean, how to transform the sensor data into user actions? In our case, we start with getting lab knowledge into our algorithms. We simulate a degradation process for the oil in artificial conditions and measure the sample, then we correlate it with sensor data.

The correlation between lab reports with diagnostics and OEM recommendations, plus sensor data leads us to create an algorithm. Other manufacturer are probably doing the same. We are adjusting values to fit the curve.

But when you are online monitoring, you are always aware of your oil condition. You can decide to change it later, you can decide to change it sooner. You don’t need to wait for the sample to decide this. You can see the trend. You can see, if a trend is a stable enough. You can choose to change the oil because of an observed trend. So the approach is different. 

Rafe Britton: If we were to set up online monitoring at a site, are we correlating the sensor data with the standard lab analysis data, with the view that at some point in the future, that site could maybe reduce the number of samples that it takes because they have now that baseline correlation between two sets of data? And that they could start to change oil based on the conditions that they see in real time?

Does that need to be done on a site by site basis that correlation, or are you doing correlations that would be applicable globally? 

Guillermo Miro: The degradations and source is heavily dependent on the lubricant. So it’s mostly a lubricant issue – so each lubricant has its own algorithm. Each oil needs a different treatment. So it’s not side-by-side but it’s oil like oil, I would say. In fact, I would say the most advanced sensors in the degradation field, they have the feature that they have several hundreds of oil references in their database, so they can cover a wide range of industries.