Machine faults can be identified by a number of symptoms such as an increase or decrease in bearing temperature or mechanical and acoustic vibration, loss of performance, increase in emissions, etc. In the case of many types of faults, it is often possible to identify an increase in mechanical vibration before other symptoms appear.
Each rotating component in a machine has an effect on the overall vibration picture. These effects can be individually identified by a dissection of the total vibration into its constituent components, to create a clearer view and allow the individual causes to be seen and identified.
The most commonly used method used for this extraction is a frequency spectrum analysis – better known as an FFT. However some faults are easier to identify using a phase analysis – e.g. misalignment - due to the fact that an FFT does not show the phase information. Other types of faults, e.g. certain gear faults, require the original time-signature or a Cepstrum analysis to be identified.
However the task of accurately diagnosing faults is nevertheless a complex one and requires intensive training, experience, powerful vibration analysers and detailed technical information about the machine.