Table of content
- What is Vibration?
- What is Vibration Analysis?
- How is Vibration Measured?
- Principles of Vibration Analysis
- Use of Vibration Analysis in Condition Monitoring
- Types of Vibration Measurement
- Applications of Vibration Analysis
- Where is Vibration Analysis Not Recommended?
What is Vibration?
An object is said to vibrate when it oscillates around a central point with a defined pattern or periodicity. Rotating components produce vibration due to centrifugal forces that can be monitored, measured, or even heard.Vibration may or may not be desirable depending on the object and situation. A pendulum’s vibration is considered essential whereas vibrations in engines are undesirable. Vibration can be initiated by electromagnetic induction, structural resonance, flow of fluids, and friction.
What is Vibration Analysis?
Vibration Analysis is the process of monitoring the patterns and levels of vibration signals. These signals are measured within a component or a machinery to detect any sort of abnormal vibration with the purpose to evaluation the condition of a test object.
How is Vibration Measured?
Measuring the displacement of the object from one point to another or the speed of displacement and acceleration are the widely used methods to quantify vibration. A vibration analysis equipment is used to capture, monitor, store, and analyse vibrations in the equipment. Advanced wireless technology, has greatly improved how data is collected and interpreted.
Modern Physics uses sensors, mounted on flat and clean surfaces, to measure speed, displacement, and/or acceleration. A piezoelectric sensor is the most widely used sensor that measures vibration by converting them into electrical charges. Microelectromechanical sensors and laser doppler vibrometers are the other two popular sensors used to measure vibration. Piezoelectric sensors generate voltages on impact and these signals are then encoded for further analysis.
Principles of Vibration Analysis
Vibration analysis works on four broad principles of time, frequency, modal and joint methods, each of which gives useful insights. The time analysis is the simplest of all and is a simple waveform monitored on the screen of an oscilloscope. While exploring the appropriate sensor, factors such as temperature and bandwidth should be considered.
In equipment with rotating parts, frequency spectrum analysis wins over time-domain analysis as the machine would generate multiple vibrations emanating from different components such as shafts, gears, etc. Frequency spectrum analysis focuses on time waveforms of every individual vibration that is significant for the study. Joint techniques such as the Gabor-Wigner-Wavelet models can be used to monitor more than one spectrum together.
Use of Vibration Analysis in Condition Monitoring
Vibration analysis analyses patterns of vibrations from the data collected by the sensors to detect any aberrations that could lead to equipment failure. Statistical methods such as Root Mean Square(RMS), Kurtosis, Standard Deviation, Degree of Skewness, etc are often employed to analyse the vibration waveforms. Every manufacturer aspires to attain market leadership by producing maximum goods at minimum cost and within a reasonable time without compromising on quality. Unscheduled maintenance, equipment failures, lead and lag times in procuring/installing replacements make preventive/predictive maintenance a necessity.
It is an excellent tool to identify faults caused due to installation errors, substandard lubrication, misalignment, etc. In the majority of the cases, faults can be detected well in advance before the fault leads to damage to the entire equipment causing significant downtime and loss of life and money.
Types of Vibration Measurement
- Holistic and General Check: This is done by manually feeling the machine to assess the level of vibration. Although this has no mathematical significance, it is an intuitive method used largely by passenger car owners.
- Spectral Analysis: Done using Fourier Transforms, Spectral Analysis identifies the peak frequency to locate the fault.
- Discrete Frequency Monitoring: This is more like a reverse check on the component in the machine. Setting the frequency to a recommended level and then monitoring associated vibrations will throw up discrepancies, if any.
- Shock Pulse Monitoring: This is used mainly for monitoring roller bearings with hand-held sensors. When roller bearings collide, vibrations are generated due to the impact. These shockwaves are used in shock pulse monitoring.
- Signal Averaging: Determining the level of signals at each frequency is important especially for lower frequency measurement as they have longer averaging times. Signal averaging finds the highest application in monitoring of gears in conjunction with their rotational speeds.
- Kurtosis: This flags spikes in random signals. Signals with higher kurtosis imply a greater number of peaks. This method finds highest applications in identifying stress in roller bearings.
- Cepstrum Analysis: This method looks at patterns in a spectrum and is used to assess gear rotation speeds and tooth meshing.
Applications of Vibration Analysis
Vibration analysis is the most favoured option for identifying faults well in advance and this is often correlated with manual data that is noted by the system technicians. This is a highly recommended method for identifying mechanical faults in DC equipment that have rotating machinery. Using a combination of both automated and manual processes gives a clearer picture of the fault lines.
For early fault detection and predictive maintenance, Spectrum Comparison is done, that compares readings observed at different times, but at the same point. To pinpoint the exact component that may be faulty, Cepstrum Analysis, of the frequency content, is recommended.
Where is Vibration Analysis Not Recommended?
For sourcing accurate data on vibrations, sensors are generally placed on or very close to the component that has to be monitored. Vibration analysis is not recommended for monitoring components that are placed in remote places such as underground shafts or in hazardous conditions that the sensors may not withstand. This method is also not recommended to identify leaks, electrical faults, and equipment placed in noisy environments.
Most sensors work up to a maximum frequency of 15kHZ. Higher frequency measurements would need a different type of sensor. Similarly, although most low frequencies are picked up, there is a strong probability that ultra-low frequencies may be ignored. Vibration analysis cannot monitor the condition of lubricants. At best it can assess whether a lubricant exists or not.
Vibrations create sounds and these sounds in turn generate pressure that again creates vibrations. A detailed study of this interrelation between sound and vibration and its impact on condition monitoring is necessary.
Although the concept of vibrations was known to scientists for more than 100 years, its applications in monitoring the performance of machines became evident only a couple of decades ago. With the advent of AI, ML & Data Analytics, vibration analysis has received a major boost in terms of accuracy in pinpointing faults well in advance thus saving billions of dollars worldwide.