Vibration analysis detects the characteristic frequency signatures of developing mechanical faults in rotating machinery — bearing defects, shaft imbalance, misalignment, looseness, and gear wear — typically weeks to months before they progress to catastrophic failure. It is the most widely implemented predictive maintenance technology globally, backed by decades of validated case studies and formalised in international standards including ISO 10816, ISO 20816, and ISO 13373. For Singapore facilities operating motors, pumps, fans, compressors, and gearboxes in demanding tropical conditions, a structured vibration programme is both a reliability and a cost-management investment.
This guide explains the physics of machine vibration, the main fault signatures and their frequencies, how to collect and analyse vibration data, what instruments are appropriate for different applications, and how to interpret results against recognised severity criteria.
The Physics of Machine Vibration
Every rotating machine vibrates. Vibration is generated by forces acting on the machine structure — imbalance in the rotating mass, misalignment between coupled shafts, rolling element contact in bearings, meshing of gear teeth, or hydraulic forces in pumps and fans. In a healthy machine, these forces are small, symmetric, and consistent; the resulting vibration is low in amplitude and stable over time.
As a machine develops a fault, additional forces are introduced. A bearing with a developing surface defect on the outer race generates a repetitive impact each time a rolling element passes over the defect. This impact occurs at a calculable frequency — the Ball Pass Frequency Outer race (BPFO) — derived from the bearing geometry and shaft speed. Spectral analysis of the vibration signal reveals this frequency as a peak in the frequency spectrum, often accompanied by sidebands at shaft speed harmonics.
This is the power of vibration analysis: each fault type has a characteristic frequency signature determined by the machine geometry. An analyst who can identify these signatures in the frequency spectrum can not only confirm that a fault is present but often identify specifically which bearing, which gear pair, or which rotor component is affected.
Common Fault Types and Their Frequency Signatures
Imbalance
Imbalance is the most common rotating machine fault. It occurs when the mass centre of the rotating assembly does not coincide with the geometric centre of rotation. The resulting vibration appears as a strong peak at 1× running speed (1X) in the radial direction. Correcting imbalance requires precision balancing — either in-place dynamic balancing or workshop balancing on a balancing machine.
Misalignment
Shaft misalignment between a motor and its driven load (pump, fan, gearbox) generates forces at 1× and 2× running speed, with axial vibration often elevated relative to radial. Angular misalignment predominantly produces axial vibration; parallel (offset) misalignment produces radial vibration at 2X. Precision laser alignment tools — available through Unitest Instruments — are used to correct misalignment to within OEM tolerances.
Bearing Defects
Rolling element bearing faults generate vibration at the four bearing defect frequencies: BPFO (outer race), BPFI (inner race), BSF (ball spin frequency), and FTF (fundamental train frequency or cage frequency). These are calculated from bearing geometry (number of rolling elements, contact angle, pitch diameter, rolling element diameter) and shaft speed. In the early stages of a bearing defect, these frequencies are best detected in the high-frequency range (often above 5,000 Hz) using enveloping or demodulation techniques. As the defect progresses, sidebands proliferate and the overall vibration level rises. Late-stage bearing failure is accompanied by broadband noise and structural resonance excitation.
Looseness
Mechanical looseness — loose foundation bolts, loose rotor fit, or clearance within a bearing — produces a rich harmonic series at shaft speed (1X, 2X, 3X, 4X, 5X…) and sometimes half-order subharmonics (0.5X, 1.5X, 2.5X). The pattern can be confused with other faults; combining vibration analysis with a physical inspection of fasteners and bearing fits usually resolves the ambiguity.
Gear Faults
Gear meshing produces vibration at the gear mesh frequency (GMF = number of teeth × shaft speed) and its harmonics. Tooth wear, pitting, or breakage modulates the gear mesh frequency with sidebands at the shaft speed of the affected gear. Distributed wear produces broadly elevated sideband amplitudes; a cracked or broken tooth produces modulation at 1× shaft speed of the defective gear.
Vibration Measurement Parameters
Vibration is measured in three parameters, each sensitive to different fault types and frequency ranges:
| Parameter | Unit | Best For | Frequency Range |
|---|---|---|---|
| Displacement | mm or µm (peak-to-peak) | Low-speed machinery, shaft orbit analysis | <10 Hz |
| Velocity | mm/s (RMS) | General machinery condition, ISO 20816 compliance | 10 Hz – 1,000 Hz |
| Acceleration | g or m/s² (peak or RMS) | High-frequency bearing defects, gearboxes | 1,000 Hz – 20,000 Hz |
ISO 20816 (the successor to ISO 10816) specifies vibration severity criteria for a wide range of machine types in terms of velocity (mm/s RMS). The standard covers induction motors, steam and gas turbines, pumps, fans, compressors, and generators. Compliance with ISO 20816 thresholds is often required for insurance, warranty, or regulatory purposes.
Instruments for Vibration Data Collection
Vibration data collection instruments range from simple single-axis pen-type meters to sophisticated route-based multi-channel analysers with built-in spectrum analysis and fault frequency databases.
For a route-based PdM programme, a handheld vibration analyser with the following capabilities is recommended:
- FFT spectrum analysis to at least 20,000 Hz (20 kHz)
- Bearing defect frequency calculation from bearing geometry or built-in bearing database
- Overall vibration level (velocity and acceleration) trending
- Route storage — the ability to pre-load the machine list and measurement point sequence and store readings against each point
- Bluetooth or USB data transfer to PdM software
- Robust enclosure suitable for industrial environments
Fluke vibration analysers distributed by Unitest Instruments meet these requirements and include integration with Fluke Connect condition monitoring software. For basic overall level monitoring at a budget, the Fluke 805 vibration meter provides a quick-check overall reading and bearing condition number suitable for operators making routine walkdown checks.
Sensor Selection and Mounting
The accelerometer (vibration sensor) is the most critical component in the measurement chain. For bearing fault detection, a sensor with a flat frequency response to at least 15 kHz is required — lower-frequency sensors will miss the high-frequency content characteristic of early-stage bearing defects.
Sensor mounting method significantly affects the usable frequency range:
- Stud mount: Direct metal-to-metal contact; highest frequency response (to full sensor specification); used for permanent monitoring points
- Magnetic mount: Quick-attach; usable to approximately 5–7 kHz; suitable for route-based monitoring at standard bearing points
- Probe or handheld: Only suitable for low-frequency overall level checks; not appropriate for bearing defect detection
For repeatable trending, measurement points should be marked on the machine housing and the same sensor mount method used at every visit. A 10% change in mounting position can produce more signal variation than a genuine 10% change in machine condition.
Calibration of Vibration Instruments
Vibration analysers and their associated accelerometers must be calibrated at defined intervals, with calibration traceable to national standards. Unitest Instruments provides SAC-SINGLAS accredited calibration across eight measurement disciplines including dimensional and physical measurements relevant to vibration instruments. For ISO 20816 compliance, calibration certificates traceability is particularly important as measurements are compared to defined numerical thresholds.
Accelerometers are sensitive devices — a dropped sensor can permanently alter its sensitivity, producing systematically incorrect amplitude readings that would corrupt the trend data on which maintenance decisions are made. Any accelerometer that has been dropped or subjected to shock should be returned for calibration before further use.
Setting Alert Thresholds and Trending
ISO 20816 provides machinery-type-specific velocity severity bands, but these are absolute thresholds for generic machine classes. For an optimised PdM programme, statistical thresholds based on your specific machine's historical baseline are more sensitive to genuine degradation and produce fewer false alarms.
A practical approach uses the machine's own baseline data (collected when the machine was in known good condition) plus two sigma (2σ) and three sigma (3σ) control limits as advisory and alarm thresholds respectively. This is analogous to statistical process control (SPC) applied to machine condition.
Bearing-specific fault frequency amplitude trending — tracking the amplitude at BPFO, BPFI, BSF, and FTF over successive measurement cycles — is more sensitive to bearing degradation than overall level trending alone, often detecting a developing defect significantly earlier. Read our practical guide to setting up a full PdM programme.
Integrating Vibration with Other PdM Technologies
Vibration analysis is most powerful in combination with other condition monitoring technologies. A bearing running hot and with elevated high-frequency vibration is a more compelling repair candidate than one showing high vibration alone — the thermal evidence confirms the vibration is not a measurement artefact. Combining vibration with oil analysis (for gearboxes and large machines with oil-lubricated bearings) provides both a mechanical fault signature and chemical evidence of wear debris in the lubricant.
For Singapore facilities with motors driving critical pumps and fans, integrating vibration monitoring with motor current analysis covers both the mechanical and electrical failure modes — a comprehensive approach that addresses the full failure mode spectrum. Contact Unitest Instruments to discuss a complete condition monitoring instrument package for your facility.