Razor Labs hones advantage at cutting edge of mining machine health

Zolotov said at IMARC SCADA systems, vibration analysis and alarms provided a sophisticated array of warnings and data to the owners and operators of heavy fixed and mobile plant and equipment. Yet in Razor Labs’ estimation up to 75% of failures happen “despite all of these systems”.

“We’ve seen that with all of the technologies companies have on site to detect problems with equipment catastrophic failures still occur, with significant financial implications,” Zolotov said.

The inability of current systems to capture failures between periodic inspections and the noise in data that masks underlying issues are two key creators of costly blind spots. Others include simple human error and “everything that cannot be covered by vibration or oil”.

“If I have a belt trip, or a pump performance issue, a pressure drop, for example, it cannot be seen in oil and vibration [analysis],” Zolotov said.

“But even where I capture an alarm, turn it into a work order and I prescribe remedial actions, if these actions are not correct, if they’re not addressing the root cause of that failure, I’m going to see that failure again and it will cause downtime.”

Razor Labs’ holistic approach uses automated root cause analysis, AI to reduce noise and visual AI to detect otherwise unseen failures. Zolotov said improving the coverage range and accuracy of failure recognition and analyses was fundamental to significantly reducing downtime and maintenance costs.

“We’ve been in this sector for more than 10 years,” he said.

“My colleagues and I have visited more than 100 sites across all major continents, really engaging with all stakeholders, from maintenance managers and superintendents to site leadership and the global COOs of organisations.

“Automated root cause diagnosis helps to focus teams on the exact actions needed to fix the issue and prevent recurrence, cutting downtime and maintenance costs.

“DataMind AI does this by fusing data from truck sensors, fluid analysis, maintenance records and even tyre information.

“In one case DataMind AI flagged anomalies in an air filter, crankcase pressure and oil filter sensors, but pinpointed the real cause as a deteriorated wiring harness on a shared 5V–ground line. The system recommended harness repair, which the team confirmed, preventing further failures that would otherwise reoccur.

“And we see this repeated again and again. Advanced root cause analysis is the key to addressing this. Don’t give me 1000s of alarms or false alarms per day. Give me five real iterations. Give me the root cause and the actions, so that when I actually complete the actions I do not see that failure again in the future.

“Mining is also very noisy and that noise is not only generating false alarms but is also masking real deteriorations.

“We can use AI to remove that noise and to compensate for all of the normal variations in mining – different payloads, different engine loads, different RPMs – to turn noisy data into a clear trend where you can see relevant phenomena a month before there is any damage.

“Now imagine the impact of this for engines, for tyres, for transmissions; preventing damage prolongs the life of the equipment and reduces very expensive replacement costs.

“And we can do this one not only for mobile fleets but also for fixed assets.”

Srulevich said visual AI added another sensory layer to the tools tracking equipment performance at mine sites.

“We’re trying to push machines to do more and by using sensor fusion technologies, by combining different sets of capabilities [and] different sources of data, we are able to understand the health of the machine better and also build a knowledge base to improve the performance of machines over time,” he said.

“We need to help customers understand more from the data that they have.

“We see the increase in AI coming together. The question we got in 2021 was, how accurate is your model? We don’t hear that anymore. Trust today is linked to your ability to deliver results that have fundamental impact within the organisation itself. Can you actually produce results that end users will leverage?

“I think this is where the market is going right now. It’s about applying AI at a level where we can demonstrate the results quickly, with a team on the ground that accumulates results and turns them into huge benefits for an organisation.”