Mechanical Vibration Analysis: Early Fault Detection

Why Vibration Analysis?

Imagine standing next to an electric motor spinning at 1500 RPM. It sounds normal to your ear, but a small sensor mounted on its housing captures invisible vibrations — frequencies and amplitudes that tell a precise story about the condition of every component inside. This is the essence of Vibration Analysis: converting mechanical motion into readable data that reveals faults weeks or months before failure.

In industrial plants — from cement factories to packaging lines — an unexpected pump or compressor failure can halt production for an entire day. Vibration analysis is the first line of defense in Predictive Maintenance, saving facilities enormous costs compared to run-to-failure strategies.

Vibration Fundamentals: Displacement, Velocity, and Acceleration

Every mechanical vibration is described by three mathematically related variables:

• Displacement: the distance a body moves from its equilibrium position. Measured in micrometers (μm) or mils (1 mil = 0.001 inch). Useful for measuring clearances and misalignment at low frequencies.
• Velocity: the rate of change of displacement over time. Measured in mm/s or in/s. The most comprehensive indicator of vibration severity because it reflects kinetic energy — and is the primary metric in international standards like ISO 10816.
• Acceleration: the rate of change of velocity. Measured in g (where 1g = 9.81 m/s²). Highly sensitive to high frequencies — ideal for detecting bearing and gear defects.

Practical rule: use displacement below 10 Hz, velocity in the 10–1000 Hz range, and acceleration above 1000 Hz.

FFT Spectrum Analysis

Imagine hearing an unusual noise from a machine — it is actually a blend of several overlapping frequencies. The time-domain waveform (amplitude versus time) does not easily reveal which frequency is causing the problem. This is where the Fast Fourier Transform (FFT) comes in.

FFT converts the signal from the time domain (time vs. amplitude) to the frequency domain (frequency vs. amplitude). The result is a spectrum showing every frequency present and its magnitude. Just as a prism separates white sunlight into the colors of the rainbow, FFT separates a complex vibration signal into its fundamental components.

Key Spectrum Terminology

• 1x: the fundamental frequency — equals the shaft rotational speed. If the shaft runs at 1500 RPM, then 1x = 25 Hz.
• 2x, 3x, …: multiples of the fundamental frequency (harmonics).
• Peak Amplitude: the height of a peak in the spectrum — indicates vibration severity at that frequency.
• Baseline: a reference spectrum taken when the machine is in good condition, used for future comparison.

Common Fault Signatures

Every mechanical fault leaves a distinctive frequency signature in the spectrum. Learning these signatures transforms you from a number reader into a skilled diagnostician.

Imbalance

The simplest and most common fault. It occurs when the center of mass does not coincide with the center of rotation — like placing a lump of clay on one side of a fan blade.

• Signature: a high peak at 1x (rotational frequency) in the radial direction.
• Characteristic: amplitude increases with the square of speed — doubling the speed quadruples the vibration.
• Remedy: in-situ dynamic balancing using a portable balancer.

Misalignment

Occurs when the motor shaft and the driven load are not collinear. Two main types:

• Angular misalignment: high peak at 1x in the axial direction.
• Parallel (offset) misalignment: high peak at 2x in the radial direction, often with harmonics at 3x and 4x.
• Remedy: laser alignment using tools such as Fixturlaser or SKF TKSA.

Bearing Defects

Defective bearings produce characteristic frequencies that depend on bearing geometry (number of balls, ball and raceway diameters):

• BPFO (Ball Pass Frequency Outer Race): outer race defect frequency. Most common because the outer race is stationary and carries the greatest load.
• BPFI (Ball Pass Frequency Inner Race): inner race defect frequency. Appears with amplitude modulation at 1x because the inner race rotates.
• BSF (Ball Spin Frequency): ball defect frequency.
• FTF (Fundamental Train Frequency): cage frequency. Very low (typically less than 0.5x).

Practical tip: you do not need to memorize the formulas — enter the bearing number (e.g., SKF 6205) into the manufacturer's database or your analysis software and it will calculate the frequencies automatically.

Gear Defects

• GMF (Gear Mesh Frequency): the number of teeth on the gear multiplied by its rotational speed. For example, a 20-tooth gear running at 30 Hz produces GMF = 600 Hz.
• Fault indicator: sidebands appearing around GMF spaced at the rotational speed of the defective gear.

Fault Signature Summary

ISO 10816 Vibration Severity Chart

ISO 10816 (and the newer ISO 20816) classifies vibration severity into four zones based on velocity (mm/s RMS) and machine class:

Machine classes: Class I (small machines up to 15 kW), Class II (medium 15–75 kW), Class III (large on rigid foundations), Class IV (large on flexible foundations such as turbines). Each class has different threshold values.

Measurement Equipment: Portable vs. Online Monitoring

Portable Collectors

A handheld device roughly the size of a smartphone with a magnetic accelerometer — place it on the bearing housing, press the measure button, and the frequency spectrum appears on screen within seconds.

• Examples: Fluke 810, SKF Microlog, Emerson CSI 2140.
• Advantages: relatively low cost, flexible, suitable for hundreds of machines.
• Disadvantages: cannot detect sudden faults between measurement rounds; depends on technician schedule adherence.

Online Monitoring Systems

Permanently installed sensors on critical machines, connected to a central monitoring system that runs around the clock.

• Examples: SKF IMx, Emerson AMS, Bently Nevada 3500.
• Advantages: instant detection of any change, automatic alerts via email or SMS, complete vibration history.
• Disadvantages: high cost (sensors + cables + system); justified only for critical equipment.

Route-Based Monitoring Programs

Equipment alone is not enough — you need a systematic program for data collection and analysis. A route-based program is the backbone of any predictive maintenance effort.

Building a Program Step by Step

1. Asset inventory: list all rotating equipment with key data (speed, power, bearing types, installation date).
2. Criticality ranking: classify machines by criticality — equipment whose failure stops production gets top priority.
3. Measurement point assignment: define measurement points on each machine (horizontal, vertical, axial on each bearing) with a consistent naming convention.
4. Interval selection: monthly for standard machines, weekly or daily for critical ones.
5. Route creation: arrange measurement points in a logical walking path to minimize travel — like planning a mail delivery route.
6. Baselines and alarm limits: take initial readings when machines are in good condition, then set alarm thresholds based on ISO 10816 or site experience.
7. Trend analysis: track values over time — a gradual rise indicates deterioration; a sudden spike signals imminent failure.

Field Tips

• Consistency: always measure from the same point, with the same sensor, under the same operating conditions (full load, stable temperature).
• Documentation: record any observations (unusual noise, heat, oil leaks) alongside the measurement — context is critical.
• Integration: correlate vibration data with thermography and oil analysis for a complete picture.
• Training: one well-trained analyst is more valuable than ten technicians collecting data without understanding.

Conclusion

Vibration analysis is not just about measuring numbers — it is the language of machines. Learning to read a frequency spectrum enables you to diagnose imbalance, misalignment, and bearing defects before they become catastrophic failures. Start with a portable collector and a disciplined route-based program, and you will see unexpected breakdowns decrease and equipment lifespans extend across your facility.