Vibration Analysis: Early Warning System for Rotating Equipment

Why Vibration Analysis Matters

Rotating equipment represents the operational backbone of most industrial facilities. Pumps, compressors, fans, turbines, and motors keep processes running. When they fail unexpectedly, the consequences ripple through the entire operation: lost production, emergency repairs, safety incidents, and downstream process upsets.

Vibration analysis is the most mature and widely applied predictive maintenance technology for rotating equipment. A well-implemented vibration monitoring program can detect developing faults weeks or months before functional failure, providing the lead time needed to plan repairs during scheduled outages rather than reacting to emergencies.

At Integral Solutions, vibration analysis is a cornerstone of the reliability programs we implement at gas plants, compressor stations, and industrial facilities across North America. This article covers the fundamentals of vibration monitoring and provides practical guidance for establishing or improving your program.

The Physics of Machine Vibration

All rotating machines vibrate. Vibration is the mechanical response to dynamic forces generated by the rotating and reciprocating components within the machine. In a healthy machine, these forces are small and well-balanced, producing low vibration levels. As faults develop, the dynamic forces change, producing characteristic changes in vibration that can be measured and analysed.

The three fundamental parameters of vibration measurement are:

Displacement measures how far the machine moves from its rest position, typically in mils (thousandths of an inch) or micrometres. Displacement is most useful for low-frequency vibration below 10 Hz and for shaft vibration measurements using proximity probes.

Velocity measures how fast the machine is moving, expressed in inches per second (in/s) or millimetres per second (mm/s). Velocity is the most commonly used parameter for general machinery monitoring because it provides a good representation of vibration severity across the frequency range of 10 Hz to 1000 Hz where most common faults manifest.

Acceleration measures the rate of change of velocity, expressed in g units. Acceleration is most sensitive to high-frequency vibration and is essential for detecting bearing defects and gear mesh problems.

Types of Vibration Measurements and Analysis

Overall Vibration Level

The simplest measurement is overall vibration, which represents the total vibration energy across all frequencies combined into a single number. This is useful for trend monitoring: if overall vibration on a pump increases from 0.15 in/s to 0.35 in/s over three months, something is changing and warrants investigation.

However, overall vibration alone cannot diagnose the cause of the change. Two machines can have identical overall vibration levels but entirely different fault conditions.

Spectral (Frequency) Analysis

Spectral analysis decomposes the overall vibration signal into its individual frequency components, revealing which specific frequencies are elevated. This is the primary diagnostic tool because different faults produce vibration at characteristic frequencies.

A machine running at 3600 RPM (60 Hz) with an unbalance condition will show dominant vibration at 1x RPM (60 Hz). The same machine with misalignment will show elevated vibration at 2x RPM (120 Hz) and potentially 3x RPM. A bearing with an outer race defect will produce vibration at the Ball Pass Frequency Outer Race, a specific frequency determined by bearing geometry and shaft speed.

By identifying which frequencies are elevated and trending their amplitudes over time, an experienced analyst can determine what is wrong, how severe it is, and how quickly it is progressing.

Time Waveform Analysis

The time waveform shows the actual vibration pattern as it occurs in time. While spectral analysis tells you which frequencies are present, the waveform reveals the character of the vibration: smooth sinusoidal patterns indicate single-frequency faults like unbalance, while impulsive patterns with sharp spikes indicate impacts from bearing defects, gear tooth damage, or looseness.

Waveform analysis is particularly valuable for detecting early-stage bearing defects, impacts and impulsive events, and truncated or clipped waveforms indicating looseness or rubs.

Envelope (Demodulation) Analysis

Bearing defect signals are often buried beneath stronger vibration from other sources. Envelope analysis applies a high-frequency bandpass filter and amplitude demodulation to extract the repetitive impact patterns produced by bearing defects, even when they are masked by other vibration sources.

This technique is essential for early bearing fault detection and is standard practice in modern vibration programs.

Common Faults and Their Vibration Signatures

Unbalance

Unbalance is the most common cause of excessive vibration in rotating machinery. It produces dominant vibration at 1x RPM in the radial direction, with amplitude proportional to the square of the speed. Unbalance vibration is characteristically smooth and sinusoidal.

Causes include mass eccentricity from manufacturing tolerances, deposit buildup on impellers or fans, missing balance weights, and non-uniform material density. Correction involves precision balancing using trial weight methods or influence coefficient balancing.

Misalignment

Shaft misalignment produces characteristic vibration patterns at 1x RPM, 2x RPM, and 3x RPM, with the 2x component often being dominant. Angular misalignment tends to produce high axial vibration, while offset misalignment produces elevated radial vibration.

High coupling temperatures, premature seal failures, and bearing failures on the coupling end of machines are common secondary symptoms. Correction requires precision alignment using laser alignment tools.

Bearing Defects

Rolling element bearing defects progress through four stages. In Stage 1, microscopic subsurface cracks produce ultrasonic emissions detectable only with high-frequency techniques. Stage 2 produces distinct bearing defect frequencies visible in envelope spectra. Stage 3 shows broadband vibration increase as defects grow and interact. Stage 4 produces dramatic vibration increase with random noise as the bearing approaches catastrophic failure.

The practical detection window is Stages 2 and 3, where defects are clearly identifiable with sufficient lead time for planned replacement. A good vibration program will detect bearing defects 2-6 months before functional failure.

Looseness

Mechanical looseness produces vibration at multiple harmonics of running speed (1x, 2x, 3x, 4x, etc.) and often includes sub-harmonics at 0.5x RPM. The time waveform is characteristically non-repeatable and impulsive.

Common causes include loose foundation bolts, cracked or broken support structures, loose bearing fits, and excessive internal clearances.

Building an Effective Vibration Program

Step 1: Equipment Selection and Prioritization

Not every machine warrants the same level of monitoring. Prioritize based on criticality (production impact of failure), consequence (safety and environmental impact), and maintainability (availability of spares and repair lead time).

Critical machines may warrant permanently installed online monitoring systems with continuous data collection and automated alerts. Important machines should be included in a periodic route-based monitoring program with data collection every 2-4 weeks. Non-critical machines may only need baseline measurements and monitoring when problems are suspected.

Step 2: Measurement Point Selection

Define consistent measurement points on each machine including bearing housings in horizontal, vertical, and axial directions. Ensure measurement points are on solid structural surfaces directly above or adjacent to bearings. Mark measurement points permanently to ensure consistency between data collections.

Step 3: Baseline Data Collection and Alert Configuration

Collect baseline data when machines are known to be in good condition. Set alert thresholds based on ISO 10816 guidelines for machinery vibration severity or, preferably, based on statistical analysis of baseline data. Two-level alerting is standard: Alert level triggers investigation, and Danger level triggers immediate action.

Step 4: Data Collection Discipline

The value of a vibration program depends entirely on consistent, quality data collection. Establish fixed routes with defined collection intervals. Train data collectors to recognize and avoid common measurement errors including poor transducer mounting, inconsistent operating conditions, and electrical interference.

Step 5: Analysis and Action

Raw data without analysis is waste. Ensure your program includes experienced analysts who review data regularly, diagnose developing faults, and provide actionable recommendations with clear severity and urgency guidance.

Return on Investment

A well-implemented vibration monitoring program typically pays for itself within 6-18 months through avoided catastrophic failures, reduced spare parts inventory through condition-based replacement timing, planned rather than emergency repairs, and extended bearing and component life through early detection and correction of root causes like misalignment and unbalance.

For a typical industrial facility with 200-500 rotating assets, the annual program cost including technology, training, and analysis support is a fraction of the cost of a single major unplanned compressor or turbine failure.