Bearings are among the most heavily loaded components in rotating machinery, and when they fail, the consequences can be costly. The challenge for engineers is that bearing damage rarely occurs suddenly; it typically develops gradually through small defects, lubrication issues, or contamination that may go unnoticed between maintenance cycles.
Vibration monitoring provides a powerful and proven method for detecting these early warning signs before failure occurs. By analysing changes in vibration behaviour — using techniques such as envelope analysis, spectral trending, and high-frequency demodulation — engineers can identify developing faults and plan maintenance proactively. In this article, we explain how bearing faults manifest as measurable vibration signals, which measurement techniques are most effective, and how the right sensor selection and monitoring platform makes the difference between early warning and unexpected failure.
How Bearing Faults Manifest as Vibration Signals?
Bearings rarely fail without warning, although the signs are frequently overlooked. Long before a breakdown occurs, subtle changes in vibration patterns begin to appear in rotating machinery. Rolling element bearings generate distinctive vibration patterns during normal operation. When defects begin to form, for example, surface pitting, cracks in the raceway, or damage to rolling elements, the interaction between these components creates additional vibration energy at specific, predictable frequences.
These frequencies are directly related to the bearing geometry and rotational speed, and are referred to as bearing defect frequencies:
-
BPFO (Ball Pass Frequency, Outer race) — indicates outer raceway damage
-
BPFI (Ball Pass Frequency, Inner race) — indicates inner raceway damage
-
BSF (Ball Spin Frequency) — indicates rolling element damage
-
FTF (Fundamental Train Frequency) — indicates cage damage
These frequencies can be calculated from the bearing's dimensional data and compared against measured vibration spectra to confirm whether a defect is developing and on which component.
Through continuous vibration monitoring, engineers can observe these emerging frequency patterns and track how they develop over time. Sensors mounted near the bearing capture vibration data that reflects the mechanical condition of the rotating assembly. As wear progresses, measurable increases in vibration amplitude or the appearance of specific fault frequencies, often accompanied by sidebands at multiples of running speed, can signal that the bearing is beginning to deteriorate.
This information allows maintenance teams to investigate the issue early, rather than waiting for visible damage or operational disruption. By identifying abnormal vibration behaviour at an early stage, engineers can plan maintenance interventions more effectively and avoid unexpected equipment downtime.
Measurement Techniques for Early Fault Detection
Bearing faults tend to first appear in the high-frequency range of the vibration spectrum, often well above the frequencies associated with imbalance or misalignment. Several measurement and analysis techniques are used to capture these signals:
-
Envelope analysis (demodulation) — the most widely used technique for early bearing fault detection. High-frequency vibration is filtered, rectified, and demodulated to reveal the underlying impact repetition rate, which can then be matched to known bearing defect frequencies.
-
Kurtosis — a statistical measure of the impulsiveness of a vibration signal. Rising kurtosis values indicate the presence of sharp, repetitive impacts consistent with developing surface defects.
-
Spike energy / HFD (High Frequency Detection) — proprietary variants of high-frequency energy measurement used by various monitoring platforms to give a single-number indicator of bearing condition.
-
Overall RMS trending — as faults develop further, broadband vibration levels increase and can be trended over time as a complementary indicator.
This is where condition monitoring analytics becomes decisive. The Sensonics SentryCMS platform analyses vibration data continuously, applying spectral analysis and alarm logic to identify patterns associated with specific bearing fault modes. By comparing measured vibration frequencies against calculated bearing defect frequencies, engineers can pinpoint the likely cause and monitor how it evolves over time — transforming raw vibration data into actionable maintenance decisions.
Sensor Selection: Accelerometers vs Proximity Probes
Accurate bearing condition data depends on selecting the right sensor type for the application. Two primary technologies are used:
Accelerometers and velocity transducers (absolute casing vibration) — mounted on the bearing housing, these sensors measure the absolute vibration of the machine casing. Piezoelectric accelerometers (such as the Sensonics PZS series) are well-suited to detecting the high-frequency signals associated with early bearing faults, while velocity transducers (PZV/VEL series) are more commonly used for lower-frequency mechanical faults such as imbalance and misalignment. Both types are available with 4–20 mA output options for direct DCS integration.
Eddy current proximity probes (relative shaft vibration) — used to measure the movement of the shaft itself relative to the bearing housing. The Sensonics XPR04 proximity probe system is an API 670 compliant solution for relative shaft vibration measurement, providing non-contact measurement of shaft displacement within the bearing clearance. This is particularly relevant for fluid film (sleeve) bearings, which are common in large turbines and generators, where shaft movement within the bearing is the primary indicator of bearing condition.
Sensor placement is as important as sensor selection. For accelerometers and velocity transducers (absolute casing vibration), transducers should be mounted as close as possible to the bearing load zone — ideally at the highest load zone, where the greatest proportion of shaft vibration transfers through the bearing housing to the sensor. Mounting surfaces must be clean, flat, and rigidly coupled to the bearing housing to avoid attenuation of high-frequency signals.
Eddy current proximity probes (relative shaft vibration) are typically mounted as X & Y pairs, positioned 90° apart and 45° either side of vertical, but critically located as close to the bearing as possible. Where eccentricity measurement is the objective rather than vibration, placement differs — ideally at the mid-point between two bearings, where shaft droop or bow is at its greatest prior to barring. As the shaft is barred and droop reduces to within acceptable eccentricity limits, the barring gear can be disengaged, and the turbine can be brought towards full steam and speed.
When combined with the Sensonics SentryCMS condition monitoring platform, continuous bearing data can be trended, alarmed, and analysed remotely — supporting predictive maintenance planning and reducing the risk of unplanned downtime on critical rotating equipment.
What Next?
Whether you are specifying sensors for a new installation, upgrading an existing vibration monitoring system, or investigating a recurring bearing fault, Sensonics can help. Our range of API 670 compliant sensors — including the PZS accelerometer series, PZV/VEL velocity transducers, and XPR04 proximity probe systems — combined with the SentryCMS condition monitoring platform, provides a complete solution for bearing health monitoring on critical rotating machinery.
Contact a Sensonics specialist to discuss your application and receive a tailored recommendation.
Bearing faults rarely fail without warning — but the signals are easy to miss without the right tools. Our latest article explains how envelope analysis, bearing defect frequencies, and the correct choice of accelerometer or proximity probe can give you advance warning of developing faults. Read the full article on the Sensonics blog.
