Mastering Machine Health: AI and Ultrasound Unlock Predictive Maintenance for Ultra-low-RPM Equipment

As industries adopt artificial intelligence (AI) technologies, predictive maintenance stands out as an effective solution to optimize machine performance and reduce costly downtime. Monitoring ultra-low-RPM machines that operate at slower speeds, however, presents some challenges. To find some answers, Machine Design spoke with Artem Kroupenev, vice president of strategy at Augury, a hardware and software provider in the predictive maintenance space working to tackle these issues with advanced technology that enables more effective monitoring and diagnostics for this class of machinery.

Ultra-low RPM machines operate at speeds as slow as 1 RPM to approximately 150 RPMs. Common across various sectors, including food production, paper manufacturing and chemicals, these machines are critical to operational efficiency; however, their slow speeds make them difficult to monitor effectively.

“Ultra-low-RPM machines are difficult to monitor continuously,” said Kroupenev, noting that the primary challenge lies in the need for high-frequency signals and long sampling periods to accurately assess machine conditions. Traditional vibration analysis techniques excel with faster rotating equipment and struggle with slow moving machinery, he says.

“This type of equipment has to have a very high frequency signal and very long sample time,” Krupnov said, complicating real-time monitoring. Conventional methods often fail to capture subtle shifts in performance, he said, which can lead to unanticipated breakdowns that can disrupt operations.

The company uses advanced machine health technologies to overcome these monitoring challenges—including AI-driven solutions and high-frequency ultrasound—to continuously monitor ultra-low-RPM equipment, providing diagnostics that traditional methods cannot achieve.

“Our solution utilizes ultrasound, a really high-frequency sample,” Kroupenev explained. He said this approach captures long samples from the machines and allows for accurate diagnostics that predict failures before they occur, claiming accuracy rates exceeding 99.9%. Interestingly, the lead time for detecting potential failures can be even longer in ultra-low-RPM machines. “A lot of issues develop slower and over time,” Kroupenev said. “This allows maintenance teams to act proactively, reducing the risk of unplanned downtime.”

Augury’s technology is not limited to detecting faults; it also provides extensive diagnostics about a range of potential issues, including lubrication problems, friction, bearing wear and mechanical looseness. The system expands the diagnostic scope beyond traditional limits, offering deeper insights into machine health. Krupnov pointed out the technology's ability to diagnose issues like “unbalance and issues with motors, slow rotating motors, rotor bars and gear friction.”

It seems the technology has delivered tangible results. “We saved over 3,000 hours of downtime for a critical piece of equipment,” he said, adding, “We see anywhere between 5- and 20-times ROI for our customers.”

Ease of Integration and Application

One major benefit of these solutions is their ease of integration into existing operational workflows. “It's a simple retrofit,” Kroupenev said. Factories can be outfitted with their sensors, minimizing operational disruption. Deployment takes only a few days, allowing companies to start monitoring their equipment without extensive downtime, he says. And the sensors can be adapted for both fast and slow rotating components, providing a comprehensive view of machine performance.

“We provide a full service end-to-end solution, which means that the deployment, maintenance and ongoing diagnostics are part of our subscription service,” he said.

Once implemented, the user interface serves as a tool for technicians and engineers, where consolidated machine health data allows for the real-time monitoring of equipment status. “We call that Machine Health 360°, he said.

The Role of AI in Predictive Maintenance

Artificial intelligence is at the core of Augury's ability to provide advanced predictive maintenance solutions. The company’s AI algorithms analyze vast amounts of machine data generated from sensors deployed across various equipment types. With machine learning, these algorithms identify patterns and anomalies that would be challenging for human analysts to detect. Kroupenev says this capability allows the company to not only diagnose issues but also predict potential failures, leading to timely interventions.

The advancements in predictive maintenance and monitoring technologies are changing not only how companies use existing machinery but also how manufacturers design new equipment. Insights gained from monitoring data can help identify common issues, which can inform improvements in future machine designs.

Kroupenev said that Augury’s AI algorithms are “crowdsourcing anonymous machine information across all the different customers.” This collective data allows original equipment manufacturers (OEMs) to refine designs and enhance durability in their products significantly. The integration of generative AI into the platform augments the decision-making process for technicians and engineers.

By providing step-by-step guidance on how to address specific issues, the AI helps standardize maintenance practices and elevates the skill sets of operational teams. “You can get insights right there and then on what are some of the steps you need to take in order to fix those issues,” Kroupenev said. This feature seems to expedite repairs while minimizing dependency on specialized knowledge, which helps with greater operational resistance in the face of unexpected mechanical challenges.