EARLY BEARING FAILURE DETECTION

Bearing failure is one of the leading causes of breakdowns in rotating equipment, often incurring costly downtime. Most rotating equipment is monitored with vibration sensors on the bearing housing for both the drive end and non-drive end. These vibration signals tend to have large amounts of noise, masking early signs of bearing failure.

Existing Methodologies

Many methods with different levels of effectiveness have been developed to detect the early warning signs of a bearing failure, enabling maintenance planners to schedule repairs and prevent unexpected failures. Some of these methods are briefly described below:

• Brush Gear Monitoring: Detect poor performance by measuring the temperature of a brush or brush holder.

• Overall Level Monitoring: Take the RMS value of the vibration level over a pre-selected bandwidth and run this value against vibration severity level standards.

• Chemical Monitoring: Insulation degradation can be monitored chemically by detecting the presence of particulate matter in the coolant gas or detecting simple and complex gases .

• Temperature Monitoring: Measure local temperature at points in the machine using embedded temperature detectors.

• Shock Pulse Monitoring: Detect defects in rolling element bearings. The condition of the bearing is assessed through a quantity known as the shock pulse value (SPV), defined as:

One of the most promising and innovative methods is to detect early bearing failure using order tracking methods, specifically The Vold-Kalman Filter. The Vold-Kalman filter is a time-domain filter that is useful for extracting order-based signals, which occur in rotating equipment situations.

The basic principles of the filter involve developing an equation to best fit the behaviour of vibration signals. If the actual value deviates from the theoretical value, an indication of failure may be seen. The benefit of the Vold-Kalman filter is how accurately the developed dynamic formula can predict the current value of the sensors while taking into account changing process conditions.

Currently Dickinson Technologies’ SciViz Studio has employed the Vold-Kalman filter to detect early bearing failure in centrifugal compressors. Sustained, large deviations out of the green areas in the graph would be a cause for concern.