AdobeStock_210003451.jpeg

Early detection of compromised and broken filters

What if you could prevent food waste in production?

What if you could stop a bad batch before it was too late?

Without Intelecy, a tiny hole in a filter resulted in 3-5 days of "lost" production as it was discovered after a lab sample was cultivated. 
With Intelecy, the customer will know this only minutes after the hole appears.

Problem: As part of the production process, the liquid is sent through various filtered to eliminate unwanted microorganisms. 
A small hole in a filter can cause unwanted bacteria flow in the process but will be difficult to detect. Therefore, lab samples are taken from all batches to ensure that the liquid is of the desired quality and bacterial balance. These bacterial samples must be cultivated, which takes 3-5 days. 
Throughout this time, production continues as usual. If the lab result returns bad, millions of liters can already have progressed in the process and have become part of the end-product. The end-product must then be withdrawn, and enormous resources of raw materials, energy consumption, and production time are wasted.

Solution: Machine learning with Intelecy Anomaly detection can assess a large volume of data to learn how a process should behave. 
It can be difficult for humans to distinguish between a "close to normal" operation and a "normal" operation. The difference can be tiny, and when humans also must monitor 200 other assets, these small deviations are easy to miss. In this case, it was particularly tricky because of several modes-of-operations, and behavior will change slightly depending on what products they are producing from the batch.

The Intelecy Anomaly detection algorithms can learn different modes-of-operation and predict the expected values with high accuracy. In this case, the algorithms have 99% accuracy. The small hole in the filter triggered a change in the process that was quickly recognized and alerted by Intelecy. Warnings from Intelecy have been integrated back to the control system and triggers an alarm at the operator screen.

Bilde fra produkt filter-case.jpg

Result: The Intelecy Anomaly models is running live, monitoring the production lines. If there is a new hole in the filter, the Intelecy Anomaly models will notify them immediately. 
They can then stop production, switch to another tank, get it washed, change the filter, and start again.

With the Anomaly-tool in Intelecy, they have optimized the process. They avoid millions of liters of liquid being wasted while avoiding wasting resources on energy, other raw materials, and factory workers' time.