Maximizing motor reliability: best practices for bearing installation and vibration monitoring

“A lot of this vibration technology is geared toward finding bearing defects because bearings take the brunt of the abuse on the equipment,” Mignano says. “It’s all transmitted through the bearings. Unbalanced alignment, looseness, resonance, the different things that can happen on a piece of equipment, all that affects the life of the bearing.”The Schaeffler OPTIME system captures seven different measurements—six of those are vibration, which each focus on a different part of the machine. OPTIME captures the following six KPIs:  

• ISO Velocity – standard alarm levels have been decided for decades
• DEMODulation – focuses on bearing defects & lubrication issues
• RMS low – a filtered acceleration reading from 2 HZ to 750 Hz
• RMS high – a filtered acceleration reading from 750 Hz to 5,000 Hz
• Kurtosis low – to capture spikiness in the waveform 2 – 750 Hz
• Kurtosis low – to capture spikiness in the waveform 750 – 5000 Hz.

“They each tell us a little bit different information about what’s going on with the machine,” Mignano says. “When we see a trend start to climb and demodulation, for example, we know something is going on with the bearing. We’re seeing some sort of metal-to-metal contact.”

Demodulation, he says, is a signal processing technique that uses filtering and waveform rectification, and vendors all have their own versions. Schaeffler’s DEMODulation trend is a demodulation measurement that can alert to different changing conditions in the bearing, such as raceway defects or lubrication issues (contamination or lack of lubrication) or metal-metal contacting. “The FFT [Fast Fourier Transform] or spectrum of the processed DEMOD waveform allows us to pinpoint the defect to the specific raceway of the bearing or even the cage or rolling elements, if they are the failing component,” Mignano says.
In general, different-sized machines have a long-standing standard for vibration velocity. Outside of those standard limits, at lower frequency vibrations, unbalance, alignment, and looseness issues will show. Mechanical stresses, hydraulic problems, and electrical defects can all contribute to premature motor failure if not properly monitored.

“Pipe strain from over tightening can put the whole motor pump train out of whack,” Mignano says.

“Cavitation is caused by a lack of liquid upstream from the pump that is cavitating. When you stand near a pump that is cavitating it can sound like there are rocks in the pumps. It means two things; the pump is not operating properly, and there is damage being caused to the pump itself,” Mignano says.

Mignano says typical industrial motors can have between 50 to 70 rotor bars inside the motor. “Rotor bars are insulated components of the motor. If the insulation starts to wear (or another defect occurs), we can see this in the vibration FFT in this range, out near 50-70 times shaft speed,” he adds.

Smart sensor networking and AI trends in predictive maintenance

As connectivity is the backbone of modern condition monitoring, Schaeffler’s OPTIME system, is built on a self-healing, data-rich network, and looking ahead, advances in artificial intelligence (AI), dynamic alarm limits, and automated reporting promise to make predictive maintenance even smarter, more adaptive, and easier to integrate into daily reliability routines.

In partnership with Wirepas, the OPTIME system uses Bluetooth low energy (BLE) mesh technology. “All of our sensors are also repeaters, so we can go from gateway to a sensor 100 meters line of sight, and from that sensor to the next sensor, we can go another 100 meters and another 100 meters. They all talk to and through each other,” Mignano says. 
The network has 40 different channels on which to send data. “It’s constantly looking at which roads are busy and which path is open, and it optimizes the data transfer in that way,” Mignano says.

“It’s a self-healing network, meaning that the sensors would automatically detect the next best path to the gateway through other sensors, and it reroutes automatically,” Kauer says. This feature is especially helpful when making changes to the system, and no setup is locked into the initial framework. “It constantly reroutes and find the most appropriate path,” Kauer says.

Schaeffler is also spending a lot of time improving its application of AI and its algorithms. Its current models can capture the speed of the shaft without a tachometer to measure. “We can pull that information out of the vibration signature,” Mignano says.

However, Schaeffler wants to push the algorithms further, looking to develop condition-based greasing, where readings from the vibration sensors can tell the lubricator when it needs to grease the bearings.

The manufacturer is also looking to develop dynamic alarm limits, for customers that make different products lines and sizes, which affect machines differently. Once the vibration signature changes with a new product, the system will automatically change the alarm limits to match that new product run.

In the future, Schaeffler will also release an AI report generator to pull information from a system’s data. “Anything you can think of to pull out of that data, it’s going to pull up a nice plot and different ways to view that information,” Mignano says.

The AI Report generator allows users to query the entire database to look for trends such as how many active sensors per plan, how many sensors with severe alarms, low batteries, and lubrication levels. The reports can be set up to automatically generate on a specific day of the week or every day and emailed to your inbox. “This allows the data to be integrated into morning or weekly reliability meetings if something concerning is occurring on a critical asset, Mignano says, and the reports are customizable to the specific user (as the plant manager typically would like a different set of parameters versus a technician or service provider).