Smarter sensors let a gearbox tell how it’s doing

A recent spinout from the University of Wyoming’s Technology Business Center has set itself the goal of improving the quality of oil monitoring in wind turbines. The company, LogiLube, says its SmartGear product provides real-time, lube oil condition monitoring, predictive analytics of lubricant’s remaining-useful-life, automated collection of in-service lube oil, and even status of the gearbox filter.

The oil flow-through device is about the size of a shoe box and mounts on the gearbox. As the oil flow through it, several sensors examine the oil’s condition. Company CEO William Gillette says the company is sensor agnostic which let them use the best available and those the client prefers.

SmartGear mounts on a gearbox and includes a purpose-built computer with controls that include A.I. and machine-learning capability. AI includes a real-time decision engine that considers the individual discrete values and applies them to an algorithm, which can trigger events, take an oil sample, or send alarms. The system is mounted to a gearbox in an NREL lab. A smartphone application lets the wind tech retrieving an oil sample simply photo the QR and barcodes to identify the sample, without writing anything down.

“The device looks for the three killers of any wind-turbine gearbox: foam, water, and particles,” says Gillette. “SmartGear auto detects on the air ratio or foam in the oil, and we do that through viscosity which is measured by several different meters or sensors. They’re typically acoustic-based. We select one over the other based on the viscosity range of the oil at the operating temperature.”

Water is detected and measured two different ways. One is by measuring the viscosity. “In fact, lab technicians have characterized each of the different oil brands used in the industry. So say it’s a particular lubricant recommended by Mobil. We would determine and measure its viscosity and dielectric constant for a range of temperatures. That data is programmed into a firmware library so we can measure and report on a temperature corrected viscosity.”

Gillette says they do the same thing for the dielectric constant. “In the lab, virgin oil is purposely contaminated with incremental parts-per-million of water, up to and beyond the condemning limit of the water contamination.”

That analysis provides the behavior of the oil and dielectric constant so when it appears again during a 30-second measurement, it is possible to identify a shift in the dielectric from water contamination or a depleting additive package. The company has built a library of oil characteristics for those lubricants used most often.

Particles are more interesting, says Gillette. “NREL, other key players, and us believe that if you can monitor, alert, and alarm on the detection and the behavioral change of wear-particle generation, you can catch an event before it turns catastrophic. It’s also important to define ‘wear’ particles. That could be the smallest detectable particle coming from a metal-to-metal contact. So we have identified abrasion and wear in a poorly lubricated or compromised scenario. The smallest wear particle detectable in real time is four microns.”

The oil cleanliness spec, ISO code 4406, lists cleanliness factors for three different particle sizes.

The chain of custody for an oil sample starts with a time recording for when the sample was collected and a signal for its retrieval and exchange for a new bottle. The chain ends the next day with the oil delivered to a lab, analyzed, and actionable information returned to the wind-farm operator with recommended action

Another monitored characteristic is the differential pressure (DP) across a gearbox filter. “Suppose a lot of water has emulsified the gearbox oil. Eventually, the filter clogs with emulsification and agglomeration of additive particles. That makes the DP skyrocket. This is not usually SCADA data. So it is important to know it happening and to correct it before the DP gets high enough to damage the filter.”

When called for, the system also extracts a sample of oil. “The system triggers the SmartGear to fill in a sample bottle with up to 500 milliliters of oil, which is sent to a laboratory where personnel will perform a Flender foam test under highly controlled lab conditions.”

Gillette adds that this is more than just oil sampling. “It’s a full on intelligent machine help system.” The system signals a tech that a sample has been taken so it can be retrieved and sent to a lab.

Lubricant manufacturers also have condemning limits on oil cleanliness. “Wear particles are the clue. We have spent close to two years vetting half a dozen different in situ detection methods from around the world for finding particle contamination. We’ve selected one, wrote firmware around it, and software to control it. It filters out air bubbles that might be considered particles. That lets the sensor count ferrous and nonferrous particles.”

The device captures data in real time and reports it out to the cloud every 30 seconds so the owner gets real-time data, up to 100 million data points a year. “So when the system fills a bottle with a sample, we know everything about the oil it holds.”

He adds that real-time condition-based monitoring, combined with fleet-wide data analytics and real-time reporting will help wind-farm operators avoid costly downtime. Maintenance plans previously based on a calendar schedule can now be tailored to an as-needed basis.

Gillette says his company is part of an NREL cooperative R&D grant that includes SKF Bearings, Siemens, Amsoil, and Flenders, which is part of Winergy. “We’re providing the onboard analytics and computing power, and in situ particle sensors,” he says.

Speak Your Mind